%WBL 27 Jul 2025 fixups by hand %gecco2025.awk Revision: 1.75 , `GECCO' Keynotes 27 Jul 2025 %gecco2025.awk Revision: 1.75 , `GECCO' Accepted-Papers 27 Jul 2025 %gecco2025.awk Revision: 1.75 , `GECCO' gecco2025_acm.bib 27 Jul 2025 @inproceedings(alonso-betanzos:2025:GECCO, author = {Amparo Alonso-Betanzos}, title = {Rethinking Efficiency in Machine Learning}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Gabriela Ochoa and Bogdan Filipic and Carlos Cotta and Gabriel Luque and Tea Tusar and Richard Allmendinger and Grant Dick and Giorgia Nadizar and Alexander Brownlee and Ahmed Kheiri and Marcella {Scoczynski Ribeiro Martins} and Hirad Assimi and Nadarajen Veerapen and Mario Andres Munoz and Christian Cintrano and Ahmed Kheiri and Jamal Toutouh and Carla Silva}, pages = {1}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {Invited keynote}, keywords = {}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3737464}, doi = {doi:10.1145/3712256.3737464}, size = {1 pages}, abstract = {The success of Artificial Intelligence (AI) has so far relied on developing increasingly precise models. However, this has come at the cost of greater complexity, requiring a higher number of parameters to estimate. As a result, model transparency and explainability have diminished, while the energy demands for training and deployment have skyrocketed. It is estimated that by 2030, AI could account for more than 30\% of the planet's total energy consumption.In this context, green and responsible AI has emerged as a promising alternative, characterized by lower carbon footprints, reduced model sizes, decreased computational complexity, and improved transparency. Various strategies can help achieve these goals, such as improving data quality, developing more energy-efficient execution models, and optimizing energy efficiency in model training and inference. These innovation approaches highlight the potential of green AI to challenge the prevailing paradigm of ever-growing models.}, notes = {GECCO-2025 A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(del-ser:2025:GECCO, author = {Javier {Del Ser}}, title = {Evolutionary Computation as a Path to Safe, Trustworthy, and Responsible General-Purpose Artificial Intelligence}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Gabriela Ochoa and Bogdan Filipic and Carlos Cotta and Gabriel Luque and Tea Tusar and Richard Allmendinger and Grant Dick and Giorgia Nadizar and Alexander Brownlee and Ahmed Kheiri and Marcella {Scoczynski Ribeiro Martins} and Hirad Assimi and Nadarajen Veerapen and Mario Andres Munoz and Christian Cintrano and Ahmed Kheiri and Jamal Toutouh and Carla Silva}, pages = {2}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {Invited keynote}, keywords = {}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3737465}, doi = {doi:10.1145/3712256.3737465}, size = {1 pages}, abstract = {As AI systems grow in capability and autonomy, concerns around safety, alignment, and trust have taken center stage. Issues such as goal misalignment, vulnerability to adversarial attacks, and the inability to generalize reliably in open-world settings are no longer theoretical: they are pressing challenges with real-world implications. At the same time, global regulatory efforts, including the EU AI Act and other emerging international frameworks, are setting strict expectations for transparency, robustness, and accountability in AI development. This keynote provides an accessible introduction to the key pillars of safe, trustworthy, responsible, and generalpurpose AI, tailored for newcomers to the field. It highlights how evolutionary computation offers a powerful, underexplored toolkit for meeting safety and trustworthy requirements. With its emphasis on diversity, adaptability, and robustness, evolutionary computation can contribute to safer learning, better generalization, and more resilient systems. The talk will bridge technical concepts with regulatory perspectives, illustrating how evolutionary approaches can help meet both the ethical and legal requirements driving the future of responsible AI systems.}, notes = {GECCO-2025 A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(schoenauer:2025:GECCO, author = {Marc Schoenauer}, title = {Evolutionary Computation: Back to the Future}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Gabriela Ochoa and Bogdan Filipic and Carlos Cotta and Gabriel Luque and Tea Tusar and Richard Allmendinger and Grant Dick and Giorgia Nadizar and Alexander Brownlee and Ahmed Kheiri and Marcella {Scoczynski Ribeiro Martins} and Hirad Assimi and Nadarajen Veerapen and Mario Andres Munoz and Christian Cintrano and Ahmed Kheiri and Jamal Toutouh and Carla Silva}, pages = {3}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {Invited keynote}, keywords = {evolutionary algorithms, autonomous search, problem representation, principles of evolutionary computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3743669}, doi = {doi:10.1145/3712256.3743669}, size = {1 pages}, abstract = {The evolution principles underlying Evolutionary Algorithms can be applied in any search space (i.e., to any representation), provided we are able to define meaningful variation operators with respect to the problem at hand. From the historical bitstring, continuous variables and Finite State Automata to advanced program or structure embeddings and beyond, EC has gradually, and sometimes painfully, earned its spurs, turning from confidential pocketknife to recognized Swiss Army Knife. I will try to illustrate this historical perspective with various examples gathered during my 35 (omg!) years of research in EC, and to demonstrate how a thorough exploitation of the past can provide useful hints for an efficient exploration of the future.}, notes = {GECCO-2025 A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(de-franca:2025:GECCO, author = {Fabricio Olivetti {de Franca} and Gabriel Kronberger}, title = {{rEGGression:} an Interactive and Agnostic Tool for the Exploration of Symbolic Regression Models}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Carola Doerr and Mike Preuss}, pages = {4--12}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, Benchmarking, Benchmarks, Software, Reproducibility}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726385}, doi = {doi:10.1145/3712256.3726385}, size = {9 pages}, abstract = {Regression analysis is used for prediction and to understand the effect of independent variables on dependent variables. Symbolic regression (SR) automates the search for non-linear regression models, delivering a set of hypotheses that balances accuracy with the possibility to understand the phenomena. Many SR implementations return a Pareto front allowing the choice of the best trade-off. However, this hides alternatives that are close to non-domination, limiting these choices. Equality graphs (e-graphs) allow to represent large sets of expressions compactly by efficiently handling duplicated parts occurring in multiple expressions. The e-graphs allow to efficiently store and query all solution candidates visited in one or multiple runs of different algorithms and open the possibility to analyze much larger sets of SR solution candidates. We introduce rEGGression, a tool using e-graphs to enable the exploration of a large set of symbolic expressions which provides querying, filtering, and pattern matching features creating an interactive experience to gain insights about SR models. The main highlight is its focus in the exploration of the building blocks found during the search that can help the experts to find insights about the studied phenomena. This is possible by exploiting the pattern matching capability of the e-graph data structure.}, notes = {GECCO-2025 BBSR A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(karns:2025:GECCO, author = {Joshua Karns and Travis Desell}, title = {Evaluation Time Bias in Asynchronous Evolutionary Algorithms: A Replication Study and a Novel Mitigation Strategy}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Carola Doerr and Mike Preuss}, pages = {13--21}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Benchmarking, Benchmarks, Software, Reproducibility}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726458}, doi = {doi:10.1145/3712256.3726458}, size = {9 pages}, abstract = {Evolutionary Algorithms (EAs) are a flexible and powerful search technique that are frequently applied to a wide variety of problems. Much of their power comes from their ease of parallelization, lending themselves well to a master-worker parallelization scheme. When a synchronous (mu, lambda)-style EA is parallelized and genome evaluation time is not constant, worker processors may spend a significant amount of time idle waiting for other genome evaluations to complete. (mu, lambda)-style EAs with a steady-state population are frequently employed to avoid this idle time. There is an existing body of work that suggests that, while this does reduce idle processor time, it may not lead to better solutions because of evaluation time bias. In this work, results from a paper that demonstrate this experimentally are fully reproduced from scratch. During this replication, the roles crossover and population initialization play in this bias were uncovered. This is evaluated experimentally and motivates a mitigation strategy, which is compared to other mitigation strategies and is found to perform about as well as other strategies, without modifying anything other than the way the population is initialized. Moreover, a new open source library was developed in order to facilitate these experiments and further investigations.}, notes = {GECCO-2025 BBSR A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(kenny:2025:GECCO, author = {Angus Kenny and Tapabrata Ray and Hemant Singh}, title = {Multi-objective L-shaped Test Functions}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Carola Doerr and Mike Preuss}, pages = {22--29}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Benchmarking, Benchmarks, Software, Reproducibility}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726323}, doi = {doi:10.1145/3712256.3726323}, size = {8 pages}, abstract = {Many real-world multi-objective optimization problems exhibit L-shaped Pareto fronts, characterized by steep trade-offs between objectives near their extreme values. This class of problems poses significant challenges for evolutionary algorithms in obtaining uniformly spread solutions across the Pareto front (PF). This paper introduces eight new test functions with L-shaped and reflected L-shaped PFs, intended to provide a valuable framework for benchmarking multi-objective optimization algorithms. The functions are based on modifications of the well-known DTLZ2 problem and a reciprocal function formulation, each formulated in standard and 'hard' variants. The 'hard' variants introduce modifications to the auxiliary functions, increasing problem difficulty by biasing the distribution of non-Pareto solutions away from the PF as the problem dimensionality increases. Numerical experiments are conducted using NSGA-II and MOEA/D algorithms, with performance evaluated using hypervolume and inverted generational distance metrics. The results demonstrate that the proposed test functions effectively challenge the algorithms, especially in their 'hard' variants, and outline the differences in algorithm performance based on the shape of the PF. These findings highlight the need for development of more robust multi-objective optimization techniques capable of handling such PF geometries.}, notes = {GECCO-2025 BBSR A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(malan:2025:GECCO, author = {Katherine Mary Malan and Mario Andres Munoz}, title = {Why We Should be Benchmarking Evolutionary Algorithms on Neural Network Training Tasks}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Carola Doerr and Mike Preuss}, pages = {30--38}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Benchmarking, Benchmarks, Software, Reproducibility}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726318}, doi = {doi:10.1145/3712256.3726318}, size = {9 pages}, abstract = {Progress in evolutionary algorithms is strongly influenced by competitions with their associated benchmark suites. These suites usually consist of artificial functions, designed to cover a range of problem complexities, such as separable and non-separable functions. For these benchmarks to be effective as proxies for fine-tuning algorithms for use in practice, they should ideally match the features found in real-world optimisation problems. The training of neural networks is an optimisation task on which evolutionary algorithms are known to perform poorly compared to gradient-based search strategies. In this paper we analyse the search spaces of a suite of neural network training tasks using exploratory landscape analysis (ELA). We show that the features of the neural network training tasks occupy a different region of ELA feature space than the widely used Black-Box Optimisation Benchmarking (BBOB) suite. We argue that continuous optimisation benchmark suites should be extended to include problems such as neural network learning tasks that exhibit weak global structure, multiple global optima and large areas of neutrality. By benchmarking on such tasks, we hope that evolutionary algorithms can be developed to provide competitive performance on the training of neural networks in scenarios where exploiting the gradient may not be the best approach.}, notes = {GECCO-2025 BBSR A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(margraf:2025:GECCO, author = {Valentin Margraf and Anna Lappe and Marcel Wever and Carolin Benjamins and Eyke Huellermeier and Marius Lindauer}, title = {{SynthACticBench:} A Capability-Based Synthetic Benchmark for Algorithm Configuration}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Carola Doerr and Mike Preuss}, pages = {39--47}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Benchmarking, Benchmarks, Software, Reproducibility}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726330}, doi = {doi:10.1145/3712256.3726330}, size = {9 pages}, abstract = {Algorithm configuration deals with the automatic optimization of an algorithm's parameters to maximize its performance on a distribution of problem instances, such as Boolean satisfiability or the traveling salesperson problem. While significant progress has been made in developing optimizers for algorithm configuration - so-called algorithm configurators - their evaluation remains computationally expensive and often relies on real-world scenarios with hard-to-control characteristics. This makes it challenging to analyze their strengths and weaknesses systematically. To address this, we introduce SynthACticBench, a synthetic benchmark specifically designed to isolate and investigate key properties of algorithm configuration problems. Our benchmark distinguishes between properties related to the configuration space and those associated with the objective function. We define a configurator's ability to handle a particular property as its capability -for example, the capability to manage hierarchical configuration spaces. Using SynthACticBench, we evaluate two state-of-the-art algorithm configurators, SMAC and irace, examining their complementary capabilities and analyzing their performances across diverse benchmark functions. By providing a controlled, scalable, and capability-based evaluation environment, SynthACticBench facilitates a more targeted analysis of algorithm configurators, helping to advance research in the field. The benchmark is available at: https://github.com/annaelisalappe/SynthACticBench/.}, notes = {GECCO-2025 BBSR A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(nisioti:2025:GECCO, author = {Eleni Nisioti and Erwan Plantec and Milton Montero and Joachim Pedersen and Sebastian Risi}, title = {When Does Neuroevolution Outcompete Reinforcement Learning in Transfer Learning Tasks?}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Carola Doerr and Mike Preuss}, pages = {48--57}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {neuroevolution, benchmarking, reinforcement learning, evolution strategies, indirect encodings, Benchmarks, Software, Reproducibility}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726475}, doi = {doi:10.1145/3712256.3726475}, size = {10 pages}, abstract = {The ability to continuously and efficiently transfer skills across tasks is a hallmark of biological intelligence and a long-standing goal in artificial systems. Reinforcement learning (RL), a dominant paradigm for learning in high-dimensional control tasks, is known to suffer from brittleness to task variations and catastrophic forgetting. Neuroevolution (NE) has recently gained attention for its robustness, scalability, and capacity to escape local optima. In this paper, we investigate an understudied dimension of NE: its transfer learning capabilities. To this end, we introduce two benchmarks: a) in stepping gates, neural networks are tasked with emulating logic circuits, with designs that emphasize modular repetition and variation b) ecorobot extends the Brax physics engine with objects such as walls and obstacles and the ability to easily switch between different robotic morphologies. Crucial in both benchmarks is the presence of a curriculum that enables evaluating skill transfer across tasks of increasing complexity. Our empirical analysis shows that NE methods vary in their transfer abilities and frequently outperform RL baselines. Our findings support the potential of NE as a foundation for building more adaptable agents and highlight future challenges for scaling NE to complex, real-world problems.}, notes = {GECCO-2025 BBSR A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(preuss:2025:GECCO, author = {Oliver Ludger Preuss and Carolin Mensendiek and Jeroen Rook and Jakob Bossek and Heike Trautmann}, title = {Automated Algorithm Configuration and Systematic Benchmarking for Heterogeneous {MNK-Landscapes}}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Carola Doerr and Mike Preuss}, pages = {58--66}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {benchmarking, multi-objective optimisation, MNK-landscapes, automated algorithm configuration, robust ranking, Benchmarks, Software, Reproducibility}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726481}, doi = {doi:10.1145/3712256.3726481}, size = {9 pages}, abstract = {MNK-landscapes are a class of multi-objective combinatorial optimisation problems that simulate interactions between system components with adjustable parameters. Recently, heterogeneous MNK-landscapes were introduced, which feature objectives with varying interdependencies, offering a new direction in multi-objective (multi-modal) landscape research. This study benchmarks various evolutionary multi-objective optimisation algorithms and a local search algorithm on such landscapes by means of automated algorithm configuration. Our systematic analysis yields various insights into the behaviour and competitiveness of these algorithms and reveals that, particularly, the omni-optimizer algorithm and iterated Pareto local search yield strong, complementary, performance. These findings facilitate the case for automated algorithm selection, which we also investigate in this paper.}, notes = {GECCO-2025 BBSR A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(sadeghi-ahouei:2025:GECCO, author = {Saba {Sadeghi Ahouei} and Aneta Neumann and Frank Neumann}, title = {Evolving Diverse Differentiating Stochastic Constraints Using Multi-objective Indicators}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Carola Doerr and Mike Preuss}, pages = {67--75}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {benchmarking, chance constraints diversity optimization, Benchmarks, Software, Reproducibility}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726377}, doi = {doi:10.1145/3712256.3726377}, size = {9 pages}, abstract = {Evolutionary diversity optimization using multi-objective indicators, aims to evolve diverse solutions considering multiple features simultaneously. In this paper, we evolve diverse discriminating instances for chance-constraint submodular problems using multi-objective quality indicators. For any pair of algorithms, discriminating instances are easy to solve by one algorithm and hard to solve by the other. These instances help with investigating the strengths and weaknesses of different algorithms in solving a given problem. Hence, the availability of diverse sets of discriminating instances for important problems is essential. We introduce a (mu + 1) evolutionary algorithm to evolve diverse differentiating instance sets for the chance-constrained maximum coverage problem. This problem contains stochastic costs on the vertices, each represented by its expected value and variance. In the selection process of our algorithm, we use inverted generational distance and hypervolume to optimize the diversity of the set. The experimental results demonstrate these indicators significantly improve the diversity of the set of instances in a multi-dimensional feature space while ensuring all of them are clearly differentiating.}, notes = {GECCO-2025 BBSR A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(seiler:2025:GECCO, author = {Moritz Vinzent Seiler and Oliver Ludger Preuss and Heike Trautmann}, title = {{RandOptGen:} A Unified Random Problem Generator for Single- and Multi-Objective Optimization Problems with Mixed-Variable Input Spaces}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Carola Doerr and Mike Preuss}, pages = {76--84}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {benchmarking, problem generator, single-objective optimization, multi-objective optimization, mixed-variable optimization, Benchmarks, Software, Reproducibility}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726478}, doi = {doi:10.1145/3712256.3726478}, size = {9 pages}, abstract = {We propose a versatile problem generator, called RandOptGen, for creating diverse and complex mixed-variable optimization problems, including single- and multi-objective problems. The generator implements a tree-based structure where decision variables from continuous, integer, and categorical domains are transformed into complex objectives using arbitrary mathematical operators. It ensures the feasibility of the generated problems through a validation process by, e.g., verifying that the objective spaces lie within predefined bounds and that multi-objective problems exhibit meaningful trade-offs, characterized by a well-formed Pareto front of the generated multi-objective problems.In our experiments, we demonstrate that our generator meaningfully extends existing benchmark sets. It allows for generating a desired number of instances that are largely diverse in existing feature spaces, including so far uncovered regions. Further, we can demonstrate that our generated instances reveal broader diversity within the performance space. We thus provide an efficient and scalable framework for generating complex and diverse optimization problems, thereby extending benchmark sets and advancing research in optimization and algorithm design.}, notes = {GECCO-2025 BBSR A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(thomson:2025:GECCO, author = {Sarah L. Thomson and Michal W. Przewozniczek}, title = {Subfunction Structure Matters: A New Perspective on Local Optima Networks}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Carola Doerr and Mike Preuss}, pages = {85--93}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Benchmarking, Benchmarks, Software, Reproducibility}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726426}, doi = {doi:10.1145/3712256.3726426}, size = {9 pages}, abstract = {Local optima networks (LONs) capture fitness landscape information. They are typically constructed in a black-box manner; information about the problem structure is not used. This also applies to the analysis of LONs: knowledge about the problem, such as interaction between variables, is not considered. We challenge this status-quo with an alternative approach: we consider how LON analysis can be improved by incorporating subfunction-based information - this can either be known a-priori or learned during search. To this end, LONs are constructed for several benchmark pseudo-boolean problems using three approaches: firstly, the standard algorithm; a second algorithm which uses deterministic grey-box crossover; and a third algorithm which selects perturbations based on learned information about variable interactions. Metrics related to subfunction changes in a LON are proposed and compared with metrics from previous literature which capture other aspects of a LON. Incorporating problem structure in LON construction and analysing it can bring enriched insight into optimisation dynamics. Such information may be crucial to understanding the difficulty of solving a given problem with state-of-the-art linkage learning optimisers. In light of the results, we suggest incorporation of problem structure as an alternative paradigm in landscape analysis for problems with known or suspected subfunction structure.}, notes = {GECCO-2025 BBSR A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(ye:2025:GECCO, author = {Huigen Ye and Yaoyang Cheng and Hua Xu and Zhiguang Cao and Hanzhang Qin}, title = {{MILPBench:} A Large-scale Benchmark Test Suite for Mixed Integer Linear Programming Problems}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Carola Doerr and Mike Preuss}, pages = {94--103}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Benchmarking, Benchmarks, Software, Reproducibility}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726324}, doi = {doi:10.1145/3712256.3726324}, size = {10 pages}, abstract = {Mixed-integer linear programming (MILP) is a cornerstone of optimization with applications across numerous domains. However, the development and evaluation of MILP-solving algorithms are hindered by existing benchmark datasets, which are often limited in scale, lack diversity, and are poorly structured, making them inadequate for systematic testing across different solving approaches, especially for machine learning (ML)-based methods. To address these issues, we introduce MILPBench, a large-scale benchmark suite comprising 100,000 MILP instances organized into 60 well-categorized classes. Using structural properties and embedding similarity metrics, we developed a novel classification framework to ensure both intra-class homogeneity and inter-class diversity. In addition to the dataset, MILPBench includes a comprehensive baseline library featuring 15 mainstream solving methods, spanning traditional solvers, heuristic algorithms, and ML-based approaches. This design enables rigorous and standardized evaluation of MILP-solving algorithms under diverse conditions. Extensive benchmarking demonstrates the utility of MILPBench as a scalable and versatile testbed for advancing MILP research, fostering innovation in solver development, and bridging the gap between optimization and machine learning.}, notes = {GECCO-2025 BBSR A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(bahlous-boldi:2025:GECCO, author = {Ryan Bahlous-Boldi and Maxence Faldor and Luca Grillotti and Hannah Janmohamed and Lisa Coiffard and Lee Spector and Antoine Cully}, title = {Dominated Novelty Search: Rethinking Local Competition in Quality-Diversity}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Emily Dolson and Mary Katherine Heinrich}, pages = {104--112}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Complex Systems}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726310}, doi = {doi:10.1145/3712256.3726310}, size = {9 pages}, abstract = {Quality-Diversity is a family of evolutionary algorithms that generate diverse, high-performing solutions through local competition principles inspired by natural evolution. While research has focused on improving specific aspects of Quality-Diversity algorithms, surprisingly little attention has been paid to investigating alternative formulations of local competition itself - a core mechanism distinguishing Quality-Diversity from traditional evolutionary algorithms. Most current approaches implement local competition using explicit collection mechanisms like fixed grids or unstructured archives. These often rely on predefined bounds or hard-to-tune parameters, presenting opportunities for alternative strategies. We outline how Quality-Diversity methods can be framed as Genetic Algorithms where local competition occurs through fitness transformations rather than explicit collection mechanisms. Inspired by this insight, we introduce Dominated Novelty Search, a Quality-Diversity algorithm that implements local competition through dynamic fitness transformations, without relying on predefined bounds or parameters. Our experiments show that Dominated Novelty Search significantly outperforms existing approaches across standard Quality-Diversity benchmarks, while maintaining its advantage in challenging scenarios like high-dimensional or unsupervised behavior spaces.}, notes = {GECCO-2025 CS A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(bayer:2025:GECCO, author = {Caleidgh Grace Bayer and Robert Smith and Malcolm Heywood}, title = {Emergent Braitenberg-style Behaviours for Navigating the {ViZDoom} 'My Way Home' Labyrinth}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Emily Dolson and Mary Katherine Heinrich}, pages = {113--121}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Complex Systems}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726331}, doi = {doi:10.1145/3712256.3726331}, size = {9 pages}, abstract = {The navigation of complex labyrinths under partially observable visual state is typically addressed using complex recurrent, convolutional learning architectures (i.e. deep reinforcement learning). Conversely, in this work, we show that navigation can be achieved through the emergent evolution of a simple Braitentberg-style vehicle. We demonstrate that the interaction between agent and labyrinth is sufficient to learn a complex navigation behaviour from simple heuristics. To do so, the approach of tangled program graphs is assumed in which programs cooperatively coevolve to develop a modular indexing scheme that employs < 2.5\% of state space. We attribute this simplicity to several biases implicit in the representation, such as: (1) the use of pixel indexing as opposed to deploying a convolutional kernel or image processing operators, and; (2) extensive support for modularity in which behaviours are always decomposed into contexts and corresponding actions.}, notes = {GECCO-2025 CS A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(coiffard:2025:GECCO, author = {Lisa Coiffard and Paul Templier and Antoine Cully}, title = {Overcoming Deceptiveness in Fitness Optimization with Unsupervised Quality-Diversity}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Emily Dolson and Mary Katherine Heinrich}, pages = {122--130}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Complex Systems}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726314}, doi = {doi:10.1145/3712256.3726314}, size = {9 pages}, abstract = {Policy optimization seeks the best solution to a control problem according to an objective or fitness function, serving as a fundamental field of engineering and research with applications in robotics. Traditional optimization methods like reinforcement learning and evolutionary algorithms struggle with deceptive fitness landscapes, where following immediate improvements leads to suboptimal solutions. Quality-diversity (QD) algorithms offer a promising approach by maintaining diverse intermediate solutions as stepping stones for escaping local optima. However, QD algorithms require domain expertise to define hand-crafted features, limiting their applicability where characterizing solution diversity remains unclear. In this paper, we show that unsupervised QD algorithms - specifically the AURORA framework, which learns features from sensory data - efficiently solve deceptive optimization problems without domain expertise. By enhancing AURORA with contrastive learning and periodic extinction events, we propose AURORA-XCon, which outperforms all traditional optimization baselines and matches, in some cases even improving by up to 34\%, the best QD baseline with domain-specific hand-crafted features. This work establishes a novel application of unsupervised QD algorithms, shifting their focus from discovering novel solutions toward traditional optimization and expanding their potential to domains where defining feature spaces poses challenges.}, notes = {GECCO-2025 CS A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(di-matteo:2025:GECCO, author = {Jacopo Michele {Di Matteo} and Oliver Weissl and Agoston Eiben}, title = {Fertility During Learning in Evolutionary Robot Systems}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Emily Dolson and Mary Katherine Heinrich}, pages = {131--139}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Complex Systems}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726382}, doi = {doi:10.1145/3712256.3726382}, size = {9 pages}, abstract = {Robot evolution systems in which bodies and brains evolve in tandem can be significantly improved by extending them with the ability to learn. Technically, this means that 'newborn' robots are given the opportunity to optimize their inherited brain to control the inherited body adequately. Robots are in an underdeveloped 'infant' stage during this learning stage since their brains and fitness are still being improved. An open issue with regard to this infancy period is that of 'fertility': Should the robot be eligible for mating during the learning stage? This paper explores two distinct approaches from the literature, based on the Triangle of Life (TOL) model, where infant robots cannot produce offspring, and the Morphological Innovation Protection (MIP) mechanism, where they can. The main contribution is a new algorithm, TOL with infant fertility (TOL+IF), inspired by MIP. Experimental comparisons with TOL and MIP show that the new method is superior. TOL+IF is successful not only in producing robots with much higher fitness but also in maintaining the population diversity at higher levels and in evolving different interesting morphologies.}, notes = {GECCO-2025 CS A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(flageat:2025:GECCO, author = {Manon Flageat and Johann Huber and Francois Helenon and Stephane Doncieux and Antoine Cully}, title = {{Extract-QD} Framework: A Generic Approach for Quality-Diversity in Noisy, Stochastic or Uncertain Domains}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Emily Dolson and Mary Katherine Heinrich}, pages = {140--148}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {quality-diversity optimisation, uncertain domains, MAP-elites, Complex Systems}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726404}, doi = {doi:10.1145/3712256.3726404}, size = {9 pages}, abstract = {Quality-Diversity (QD) has demonstrated potential in discovering collections of diverse solutions to optimisation problems. Originally designed for deterministic environments, QD has been extended to noisy, stochastic, or uncertain domains through various Uncertain-QD (UQD) methods. However, the large number of UQD methods, each with unique constraints, makes selecting the most suitable one challenging. To remedy this situation, we present two contributions: first, the Extract-QD Framework (EQD Framework), and second, Extract-MAP-Elites (EME), a new method derived from it. The EQD Framework unifies existing approaches within a modular view, and facilitates developing novel methods by interchanging modules. We use it to derive EME, a novel method that consistently outperforms or matches the best existing methods on standard benchmarks, while previous methods show varying performance. In a second experiment, we show how our EQD Framework can be used to augment existing QD algorithms and, in particular, the well-established Policy-Gradient-Assisted-ME method, and demonstrate improved performance in uncertain domains at no additional evaluation cost. For any new uncertain task, our contributions now provide EME as a reliable "first guess" method, and the EQD Framework as a tool for developing task-specific approaches. Together, these contributions aim to lower the cost of adopting UQD insights in QD applications.}, notes = {GECCO-2025 CS A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(janmohamed:2025:GECCO, author = {Hannah Janmohamed and Antoine Cully}, title = {Multi-Objective Quality-Diversity in Unstructured and Unbounded Spaces}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Emily Dolson and Mary Katherine Heinrich}, pages = {149--157}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Complex Systems}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726394}, doi = {doi:10.1145/3712256.3726394}, size = {9 pages}, abstract = {Quality-Diversity algorithms are powerful tools for discovering diverse, high-performing solutions. Recently, Multi-Objective Quality-Diversity (MOQD) extends QD to problems with several objectives while preserving solution diversity. MOQD has shown promise in fields such as robotics and materials science, where finding tradeoffs between competing objectives like energy efficiency and speed, or material properties is essential. However, existing methods in MOQD rely on tessellating the feature space into a grid structure, which prevents their application in domains where feature spaces are unknown or must be learned, such as complex biological systems or latent exploration tasks. In this work, we introduce Multi-Objective Unstructured Repertoire for Quality-Diversity (MOUR-QD), a MOQD algorithm designed for unstructured and unbounded feature spaces. We evaluate MOUR-QD on five robotic tasks. Importantly, we show that our method excels in tasks where features must be learned, paving the way for applying MOQD to unsupervised domains. We also demonstrate that MOUR-QD is advantageous in domains with unbounded feature spaces, outperforming existing grid-based methods. Finally, we demonstrate that MOUR-QD is competitive with established MOQD methods on existing MOQD tasks and achieves double the MOQD-SCORE in some environments. MOUR-QD opens up new opportunities for MOQD in domains like protein design and image generation.}, notes = {GECCO-2025 CS A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(mertan:2025:GECCO, author = {Alican Mertan and Nick Cheney}, title = {Controller Distillation Reduces Fragile Brain-Body Co-Adaptation and Enables Migrations in {MAP-Elites}}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Emily Dolson and Mary Katherine Heinrich}, pages = {158--166}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {evolutionary robotics, soft robotics, brain-body co-optimization, quality-diversity, MAP-elites, Complex Systems}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726368}, doi = {doi:10.1145/3712256.3726368}, size = {9 pages}, abstract = {Brain-body co-optimization suffers from fragile co-adaptation where brains become over-specialized for particular bodies, hindering their ability to transfer well to others. Evolutionary algorithms tend to discard such low-performing solutions, eliminating promising morphologies. Previous work considered applying MAP-Elites, where niche descriptors are based on morphological features, to promote better search over morphology space. In this work, we show that this approach still suffers from fragile co-adaptation: where a core mechanism of MAP-Elites, creating stepping stones through solutions that migrate from one niche to another, is disrupted. We suggest that this disruption occurs because the body mutations that move an offspring to a new morphological niche break the robots' fragile brain-body co-adaptation and thus significantly decrease the performance of those potential solutions - reducing their likelihood of outcompeting an existing elite in that new niche. We use a technique, we call Pollination, that periodically replaces the controllers of certain solutions with a distilled controller with better generalization across morphologies to reduce fragile brain-body co-adaptation and thus promote MAP-Elites migrations. Pollination increases the success of body mutations and the number of migrations, resulting in better quality-diversity metrics. We believe we develop important insights that could apply to other domains where MAP-Elites is used.}, notes = {GECCO-2025 CS A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(schroeder:2025:GECCO, author = {Gijs Schroeder and Johannes Textor}, title = {Classifier Systems as Linear Probability Models}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Emily Dolson and Mary Katherine Heinrich}, pages = {167--175}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Complex Systems}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726453}, doi = {doi:10.1145/3712256.3726453}, size = {9 pages}, abstract = {Classifier systems solve regression and classification problems in high-dimensional spaces by generating and evolving large populations of simple rules; examples include learning classifier systems and artificial immune systems. The properties of these flexible adaptable systems are less well understood than those of more classical machine learning algorithms. Here, we reveal a deep connection between classifier systems and probabilistic models such as Na\"{\i}ve Bayes and Markov chains by showing that all of these can be expressed as generalized linear probability models. This connection shows that any probability distribution can in principle be expressed by a classifier system. We then harness this new perspective to investigate the tradeoff between model complexity and calibration - i.e., the ability to accurately fit the sequence probabilities observed in the training set - for classifier systems applied to sequence probability modeling. Contrasting our results to Markov chains of varying order, we find that a simple model classifier system has a broadly similar complexity-calibration tradeoff. We hope that our approach paves the way for further systematic investigation of the fundamental properties of classifier systems, which could make them more accessible for the machine learning community.}, notes = {GECCO-2025 CS A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(bouamama:2025:GECCO, author = {Salim Bouamama and Christian Blum}, title = {Application of {PBIG} to the Minimum Global Domination Problem}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {176--183}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {minimum global domination, population-based iterated greedy algorithm, semi-greedy solution construction, graph theory, Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726376}, doi = {doi:10.1145/3712256.3726376}, size = {8 pages}, abstract = {The Minimum Global Domination (MGD) problem is a challenging NP-hard variant of the classical Minimum Dominating Set (MDS) problem, which has numerous practical applications. Given an undirected graph, a global dominating set is a set of vertices that dominates all vertices both in the given graph and in its complement graph. In this work, we propose a Population-Based Iterated Greedy (PBIG) algorithm to effectively address the MGD problem. The algorithm employs a semi-greedy solution reconstruction strategy and a redundancy removal mechanism to enhance efficiency and solution quality. We benchmark PBIG against current state-of-the-art approaches, including the CMSA metaheuristic and the CPLEX solver, across 1440 problem instances. Experimental results demonstrate that PBIG outperforms existing methods in solution quality while significantly reducing computational time, establishing it as a powerful and efficient algorithm for the MGD problem.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(da-ros:2025:GECCO, author = {Francesca {Da Ros} and Luca {Di Gaspero} and Lucas Kletzander and Marie-Louise Lackner and Nysret Musliu and Andrea Schaerf}, title = {Dynamic Temperature Control of Simulated Annealing using Hyper-Heuristics}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {184--194}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726390}, doi = {doi:10.1145/3712256.3726390}, size = {11 pages}, abstract = {This paper explores the potential for dynamically adapting the temperature of Simulated Annealing (SA) in a problem-independent manner, eliminating the need for extensive tuning or prior knowledge of instance-specific features. Our goals are to bypass expensive tuning procedures and to ensure a balanced interplay between exploration and exploitation at appropriate stages of the search process. To achieve this, we developed a framework called HHSA that employs Hyper-Heuristics (HHs) and makes use of fixed-temperature SA as their low-level heuristics. The proposed approach is evaluated across three state-of-the-art HHs and four problem domains (i.e., k-Graph Coloring, Permutation Flowshop, Traveling Salesperson, and Facility Location). Comparative results against a fine-tuned SA reveal that HHSA consistently achieves comparable or superior results in three out of the four studied problems. The findings reinforce the broader applicability of hyper-heuristics, demonstrating their potential to generalize across different problem domains without relying on instance-specific configurations.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(dong:2025:GECCO, author = {Yun Dong and Lixin Tang and Weiyan Jia}, title = {A Learning-assisted Discrete Differential Evolution for Resource Constrained Project Scheduling}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {195--203}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726397}, doi = {doi:10.1145/3712256.3726397}, size = {9 pages}, abstract = {This paper studies a resource-constrained project scheduling problem, aiming to optimize the start times of project activities under resource and precedence constraints in order to minimize the makespan. To solve this complex problem more efficiently, we propose a problem-specific solution algorithm that combines hybrid metaheuristics with machine learning techniques. Specifically, a discrete differential evolution serves as the main framework, which is augmented with adaptive mutation, crossover, and parameter strategies. During the evolution phase, the differential evolution competes with a sequential pattern-based adaptive large-neighborhood search to generate offspring solutions. In the subsequent selection phase, a precedence decomposition scheme cooperates with a Hamming distance-based k-nearest neighbor model to evaluate the offspring solutions. Extensive experiments using benchmark datasets demonstrate that each algorithmic component positively contributes to performance enhancement, and the proposed algorithm outperforms state-of-the-art algorithms. Furthermore, we analyze the search behavior of the algorithm from various views to assess the influence of different strategies on its overall performance.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(fieldsend:2025:GECCO, author = {Jonathan Fieldsend and Arnaud Liefooghe and Katherine Malan and Sebastien Verel}, title = {Local Optima Networks for Constrained Search Spaces}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {204--212}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726303}, doi = {doi:10.1145/3712256.3726303}, size = {9 pages}, abstract = {Local Optima Networks (LON)s have been used extensively to understand the global structure of optimisation problems and to study algorithm behaviour. The central idea is to compress the search space into a graph object capturing the local optima along with information on their basins of attraction and the connections between them. This enables the visualisation of high dimensional search spaces, and the extraction of metrics for characterising and contrasting different problem instances. In this paper we extend the canonical LON definition to encompass search spaces with constraints. We use a well-known pairwise comparison operator for constrained problems, and capture the features of the constraint violation landscape that present a challenge for such an operator, such as infeasible local traps. The concept of a constrained LON is illustrated through a range of problem instances. Most problems in the context of real-world applications have constraints. By including the notion of feasibility and constraint violation into the definition of LONs, it becomes possible to use this powerful analysis tool on a much wider range of real-world problems.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(gjergji:2025:GECCO, author = {Ida Gjergji and Lucas Kletzander and Nysret Musliu and Andrea Schaerf}, title = {Large Neighborhood Search for Capacitated Facility Location with Customer Incompatibilities}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {213--221}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {large neighborhood search, facility location problem, discrete optimization, Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726355}, doi = {doi:10.1145/3712256.3726355}, size = {9 pages}, abstract = {A new variant of the classic capacitated facility location problem, which considers incompatibilities between customers, has recently been introduced in the literature. This problem captures the situation where given pairs of customers cannot be served by the same facility. Such a feature is crucial for many practical cases of location problems, such as the presence of hazardous or polluting materials or contention between competing costumers. In this paper, we propose a Large Neighborhood Search (LNS) method to solve this problem. Within the framework of LNS, we introduce three different destroy operators and we use an exact solver in the repair phase. We critically analyze the effectiveness and the efficiency of both destroy and repair operators. The experimental analysis shows that our new method outperforms existing state-of-the-art metaheuristics, providing new best solutions for all available benchmark instances.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(gong:2025:GECCO, author = {Cheng Gong and Ping Guo and Lie Meng Pang and Qingfu Zhang and Hisao Ishibuchi}, title = {Performance Comparison between Evolutionary Algorithms and Linear Programming-based Relaxation Methods for Multi-Objective Knapsack Problems}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {222--230}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {multi-objective optimization, large-scale combinatorial optimization, evolutionary computation, EMO algorithms, Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726379}, doi = {doi:10.1145/3712256.3726379}, size = {9 pages}, abstract = {Recently, performance comparison results of evolutionary multi-objective optimization (EMO) algorithms have been reported in many studies. However, EMO algorithms have not been compared with mathematical programming-based methods in those studies. To demonstrate the usefulness of EMO algorithms, it is needed to clearly show their advantages over mathematical programming-based methods in solving multi-objective optimization problems since those methods are usually highly efficient and effective. In this paper, a novel improved linear programming-based relaxation method, named ILP-R, is proposed for addressing multi-objective knapsack problems (MOKP), which are used as the test problems for performance comparison. Extensive experimental results show that ILP-R outperforms a basic linear programming-based relaxation method and EMO algorithms. Nevertheless, EMO algorithms exhibit the ability to further improve the solutions generated by the ILP-R method. Furthermore, a knowledge-based mutation method is explored to demonstrate its effectiveness in further improving the performance of EMO algorithms that use the heuristic ILP-R solutions as the initial population.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(heins:2025:GECCO, author = {Jonathan Heins and Darrell Whitley and Pascal Kerschke}, title = {To Repair or Not to Repair? Investigating the Importance of {AB-Cycles} for the State-of-the-Art {TSP} Heuristic {EAX}}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {231--239}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726436}, doi = {doi:10.1145/3712256.3726436}, size = {9 pages}, abstract = {The Edge Assembly Crossover (EAX) algorithm is the state-of-the-art heuristic for solving the Traveling Salesperson Problem (TSP). It regularly outperforms other methods, such as the Lin-Kernighan-Helsgaun heuristic (LKH), across diverse sets of TSP instances. Essentially, EAX employs a two-stage mechanism that focuses on improving the current solutions, first, at the local and, subsequently, at the global level. Although the second phase of the algorithm has been thoroughly studied, configured, and refined in the past, in particular, its first stage has hardly been examined.In this paper, we thus focus on the first stage of EAX and introduce a novel method that quickly verifies whether the AB-cycles, generated during its internal optimization procedure, yield valid tours - or whether they need to be repaired. Knowledge of the latter is also particularly relevant before applying other powerful crossover operators such as the Generalized Partition Crossover (GPX). Based on our insights, we propose and evaluate several improved versions of EAX. According to our benchmark study across 10 000 different TSP instances, the most promising of our proposed EAX variants demonstrates improved computational efficiency and solution quality on previously rather difficult instances compared to the current state-of-the-art EAX algorithm.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(jardee:2025:GECCO, author = {William Jardee and John Sheppard}, title = {Ant Colony Optimization with Policy Gradients and Replay}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {240--248}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726452}, doi = {doi:10.1145/3712256.3726452}, size = {9 pages}, abstract = {Ant Colony Optimization (ACO) has served as a widely used metaheuristic algorithm for decades for solving combinatorial optimization problems. Since its initial construction, ACO has seen a wide variety of modifications and connections to Reinforcement Learning (RL). Substantial parallels can be seen as early as 1995 with Ant-Q's relationship with Q-learning, through 2022 with ADACO's connection with Policy Gradient. In this work, we describe ACO, more specifically the Stochastic Gradient Descent ACO algorithm (ACOSGD), explicitly as an off-policy Policy Gradient (PG) method. We also incorporate experience replay into several ACO algorithm variants, including AS, MaxMin-ACO, ACOSGD, ADACO, and our two policy gradient-based versions: PGACO and PPOACO, drawing the connection to elitist ACO strategies. We show that our implementation of PG in ACO with experience replay and a baselined reward update strategy applied to eight TSP problems of varying sizes performs competitively with both fundamental ACO and SGD-based ACO versions. We also show that the replay buffer seems to unilaterally improve the performance of ACO algorithms through an ablation study.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(martynov:2025:GECCO, author = {Pavel Martynov and Maxim Buzdalov and Sergey Pankratov and Vitaliy Aksenov and Stefan Schmid}, title = {In the Search of Optimal Tree Networks: Hardness and Heuristics}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {249--257}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726425}, doi = {doi:10.1145/3712256.3726425}, size = {9 pages}, abstract = {Traffic in datacenters may follow some pattern: some pairs of servers communicate more frequently than others. Demand-oblivious networks may perform poorly for such workloads, and demand-aware networks optimized for traffic should be used instead. Unfortunately, not all shapes of networks are feasible in real hardware. Practical limitations are usually provided in the form of a topology. For example, a network may be required to be a binary tree, a bounded-degree graph or a Fat tree.In this work, we consider a topology of a binary tree, one of the most fundamental network topologies. We show that already finding an optimal demand-aware binary tree network is NP-hard. Then, we explore how various optimization techniques, including simple local searches, as well as deterministic mutation and crossover operators, cope with generating efficient tree networks on real-life and synthetic workloads.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(matsuo:2025:GECCO, author = {Takumi Matsuo and Kento Uchida and Shinichi Shirakawa}, title = {Elitist Evolutionary Algorithm for Optimization on Sets of Points}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {258--266}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {discrete optimization, elitist strategy, graph theory, Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726472}, doi = {doi:10.1145/3712256.3726472}, size = {9 pages}, abstract = {This study focuses on the search space composed of disjoint sub-spaces, each containing common or distinct finite points on Euclidean space. This problem setting is called an optimization problem on sets of points (SoP), and acceptable solutions are constructed by selecting possible points in the subspaces. In optimization on SoP, it is essential to capture the positional relation between the points. Recently, CMA-ES-SoP was proposed as an optimization method on SoP by introducing additional mechanisms based on the Delaunay diagram to CMA-ES. However, there are two problems: the worst-case complexity of the Delaunay diagram is exponential in the number of dimensions, and the convergence of CMA-ES-SoP is relatively slow because of the non-elitist strategy. In this study, we propose an elitist evolutionary algorithm for the optimization on SoP. The proposed method, (1+1)-EA-SoP, adaptively switches two mutation methods; the neighboring-point mutation selects the mutated point from the neighbors on the graph, and the global mutation randomly selects one point. In addition, we develop a novel graph structure that can be constructed with polynomial complexity and possesses several desirable properties related to the Delaunay diagram. The experimental results show that (1+1)-EA-SoP with the proposed graph realizes an effective optimization on SoP.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(pang:2025:GECCO, author = {Junwei Pang and Yi Mei and Mengjie Zhang}, title = {A Multiform Many-Objective Genetic Programming Method for Dynamic Flexible Job Shop Scheduling}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {267--276}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, dynamic flexible job shop scheduling, many-objective optimisation, multiform optimisation, Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726371}, doi = {doi:10.1145/3712256.3726371}, size = {10 pages}, abstract = {Genetic programming-based hyper-heuristic approaches have successfully evolved effective scheduling heuristics for dynamic flexible job shop scheduling. However, in addition to effectiveness, users may prefer other important factors such as model size (i.e., bloat control), structural complexity, and interpretability. To evolve scheduling heuristics considering a wide range of factors, we aim to solve a new many-objective optimisation problem with one effectiveness indicator and four commonly considered model structural complexity measures. To solve this problem, we design a new multiform many-objective genetic programming-based hyper-heuristic algorithm, which optimises this proposed many-objective optimisation task and a constructed single-objective auxiliary task in a multitask manner. This auxiliary task is specifically designed to optimise effectiveness, aiming to find effective individuals and provide beneficial genetic materials for the original task to improve search performance via knowledge transfer. The experimental results show that this approach can produce scheduling heuristics that approximate the Pareto front better than the compared state-of-the-art algorithms across a series of scenarios. Further analysis demonstrates the interpretability of evolved scheduling heuristics and the advantages of considering comprehensive structural complexity measures simultaneously.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(phan-duc:2025:GECCO, author = {Hung {Phan Duc} and Duc {Bui Trong} and Tam {Nguyen Thi} and Binh {Huynh Thi Thanh}}, title = {Pareto Front Grid Guided Multiobjective Optimization In Dynamic Pickup And Delivery Problem Considering Two-Sided Fairness}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {277--285}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {fairness, multi-objective optimization, multi-objective algorithms, dynamic pickup and delivery problem, Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726467}, doi = {doi:10.1145/3712256.3726467}, size = {9 pages}, abstract = {The dynamic delivery problem poses a complex challenge with many practical applications in logistics and transportation. Unlike the static delivery problem, where all order details are known, the dynamic delivery problem deals with continuously evolving information, with only partial data about orders available at any given moment. This paper presents the Multi-objective Dynamic Pickup and Delivery Problem with Time Windows framework, which integrates multiple objectives, including minimizing energy consumption, reducing waiting time, and ensuring fairness for both customers and vehicles. The goal is to lower overall system costs while balancing the customer experience and the workload of service providers. Previous research has primarily focused on optimizing a single objective or converting other objectives into constraints, which can limit the flexibility and effectiveness of the solutions. Our approach tackles this challenge by introducing a Pareto Front Grid-guided Multi-Objective Evolutionary Algorithm that incorporates two-sided fairness-ensuring equitable treatment for customers and service providers. The experimental results reveal that our method substantially outperforms existing multi-objective and single-objective algorithms specifically designed for the dynamic pickup and delivery problem on Hypervolume metric and Inverted Generational Distance metric.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(pi\k{a}tek:2025:GECCO, author = {Jundefineddrzej Pi\k{a}tek and Michal Witold Przewozniczek and Francisco Chicano and Renato Tinos}, title = {Moving between high-quality optima using multi-satisfiability characteristics in hard-to-solve {Max3Sat} instances}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {286--294}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726428}, doi = {doi:10.1145/3712256.3726428}, size = {9 pages}, abstract = {Gray-box optimization proposes effective and efficient optimizers of general use. To this end, it leverages information about variable dependencies and the subfunction-based problem representation. These approaches were already shown effective by enabling tunnelling between local optima even if these moves require the modification of many dependent variables. Tunnelling is useful in solving the maximum satisfiability problem (MaxSat), which can be reformulated to Max3Sat. Since many real-world problems can be brought to solving the MaxSat/Max3Sat instances, it is important to solve them effectively and efficiently. Therefore, we focus on Max3Sat instances for which tunnelling fails to introduce improving moves between locally optimal high-quality solutions and the region of globally optimal solutions. We analyze the features of such instances on the ground of phase transitions. Based on these observations, we propose manipulating clause-satisfiability characteristics that allow connecting high-quality solutions distant in the solution space. We use multi-satisfiability characteristics in the optimizer built from typical gray-box mechanisms. The experimental study shows that the proposed optimizer can solve those Max3Sat instances that are out of the grasp of state-of-the-art gray-box optimizers. At the same time, it remains effective for instances that have already been successfully solved by gray-box.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(przewozniczek:2025:GECCO, author = {Michal Witold Przewozniczek and Francisco Chicano and Renato Tinos and Jakub Nalepa and Bogdan Ruszczak and Agata Wijata}, title = {On Revealing the Hidden Problem Structure in Real-World and Theoretical Problems Using Walsh Coefficient Influence}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {295--303}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {variable dependency, gray-box optimization, genetic algorithms, dependency strength, optimization, walsh decomposition, Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726410}, doi = {doi:10.1145/3712256.3726410}, size = {9 pages}, abstract = {Gray-box optimization employs Walsh decomposition to obtain non-linear variable dependencies and use them to propose masks of variables that have a joint non-linear influence on fitness value. These masks significantly improve the effectiveness of variation operators. In some problems, all variables are non-linearly dependent, making the aforementioned masks useless. We analyze the features of the real-world instances of such problems and show that many of their dependencies may have noise-like origins. Such noise-caused dependencies are irrelevant to the optimization process and can be ignored. To identify them, we propose extending the use of Walsh decomposition by measuring variable dependency strength that allows the construction of the weighted dynamic Variable Interaction Graph (wdVIG). wdVIGs adjust the dependency strength to mixed individuals. They allow the filtering of irrelevant dependencies and re-enable using dependency-based masks by variation operators. We verify the wdVIG potential on a large benchmark suite. For problems with noise, the wdVIG masks can improve the optimizer's effectiveness. If all dependencies are relevant for the optimization, i.e., the problem is not noised, the influence of wdVIG masks is similar to that of state-of-the-art structures of this kind.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(santos:2025:GECCO, author = {Daniela Santos and Kathrin Klamroth and Pedro Martins and Luis Paquete}, title = {A Path-Relinking-based Heuristic for the Multiobjective Subgraph Problem}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {304--312}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726386}, doi = {doi:10.1145/3712256.3726386}, size = {9 pages}, abstract = {Given a simple undirected graph G, the Multiobjective Subgraph (MOS) problem aims to find a subgraph in G that maximizes the number of edges while minimizing the number of vertices. Addressing the MOS problem allows to solve the related Multiobjective Quasi-clique problem, which seeks a quasi-clique with maximum density and number of vertices and has many real-life applications. These problems have only been addressed using exact methods, which can be computationally intensive due to their NP-hard nature. In this paper, we introduce a heuristic method for solving the MOS problem. We show that a subset of optimal MOS subgraphs exhibits a nestedness property, meaning they satisfy an inclusion-wise relation. We explore this property to develop a path-relinking-based heuristic, where subgraphs from this subset serve as starting and ending points of a path to find new high-quality subgraphs. Additionally, we derive an upper bound on the number of edges for MOS subgraphs, which is used to evaluate the quality of the subgraphs generated by our heuristic. Experimental results on synthetic and real-life sparse graphs indicate that our heuristic produces high-quality subgraphs, with an average error of 2.3 edges compared to the exact method, while spending only 6.2\% of its runtime.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(santucci:2025:GECCO, author = {Valentino Santucci and Marco Baioletti and Marco Tomassini}, title = {Smooth Transition Instance Chains in Combinatorial Optimization Problems}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {313--321}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {combinatorial optimization, instance space, smooth transition, maximum cut problem, Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726351}, doi = {doi:10.1145/3712256.3726351}, size = {9 pages}, abstract = {In this work, by using an adiabatic principle and the Maximum Cut Problem, we investigate the evolution of problem instances from a given initial instance to a given final instance. The path followed goes from one instance to the next by using a statistical concept of distance such that the transition is smooth in the sense that this distance is short. In other words, the process takes place in the instance space by following a trajectory of minimal change. During the process we study the evolution of the similarity between consecutive instances and the movement of the global optima. In particular, we investigated whether a smooth path in the instance space always exists between the initial and the final instance. This allow us to discuss a number of statistical results that are of general interest for the understanding of the instance space of difficult combinatorial optimization problems.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(tan:2025:GECCO, author = {Leshan Tan and Chenwei Jin and Xinan Chen and Rong Qu and Ruibin Bai}, title = {{PGU-SGP:} A Pheno-Geno Unified Surrogate Genetic Programming For Real-life Container Terminal Truck Scheduling}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {322--330}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726326}, doi = {doi:10.1145/3712256.3726326}, size = {9 pages}, abstract = {Data-driven genetic programming (GP) has proven highly effective in solving combinatorial optimization problems under dynamic and uncertain environments. A central challenge lies in fast fitness evaluations on large training datasets, especially for complex real-world problems involving time-consuming simulations. Surrogate models, like phenotypic characterization (PC)-based K-nearest neighbors (KNN), have been applied to reduce computational cost. However, the PC-based similarity measure is confined to behavioral characteristics, overlooking genotypic differences, which can limit surrogate quality and impair performance. To address these issues, this paper proposes a pheno-geno unified surrogate GP algorithm, PGU-SGP, integrating phenotypic and genotypic characterization (GC) to enhance surrogate sample selection and fitness prediction. A novel unified similarity metric combining PC and GC distances is proposed, along with an effective and efficient GC representation. Experimental results of a real-life vehicle scheduling problem demonstrate that PGU-SGP reduces training time by approximately 76\% while achieving comparable performance to traditional GP. With the same training time, PGU-SGP significantly outperforms traditional GP and the state-of-the-art algorithm on most datasets. Additionally, PGU-SGP shows faster convergence and improved surrogate quality by maintaining accurate fitness rankings and appropriate selection pressure, further validating its effectiveness.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(tasnadi:2025:GECCO, author = {Zoltan Tasnadi and Balazs Vass and Noemi Gasko}, title = {Ant Colony Optimization Algorithm for Safest Path Computation in Presence of Correlated Failures in Backbone Networks}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {331--339}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {safest path problem, correlated failures, ant colony optimization, Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726346}, doi = {doi:10.1145/3712256.3726346}, size = {9 pages}, abstract = {Safest path computation with multiple correlated failures is a challenging computational task, with several application possibilities. In communication backbone networks, for example, establishing a path as safe as possible between the two communication endpoints is a crucial component for achieving the ambitious availability requirements on which emerging technologies like autonomous driving, AR/VR applications, or telesurgery depend. In this paper, after proving the NP-hardness of the problem, we propose the Safest Path Ant Colony Optimization (SP-ACO) algorithm to solve the problem. The proposed algorithm is based on the Max-Min Ant System. Numerical experiments conducted on both real-world and synthetic inputs prove the effectiveness of the proposed approach. The proposed SP-ACO algorithm typically provides at least as safe paths as the state-of-the-art algorithms, even outperforming them in a significant share of the parameter settings. This grants a place for the SP-ACO among the best solutions for safest path finding in the presence of correlated failures.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(wittner:2025:GECCO, author = {Dominic Wittner and Jakob Bossek}, title = {Cluster Prevention in Evolutionary Diversity Optimization for Parallel Machine Scheduling}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {340--348}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {evolutionary diversity optimization, diversity measurement, parallel machine scheduling, Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726357}, doi = {doi:10.1145/3712256.3726357}, size = {9 pages}, abstract = {This paper addresses the prevention of undesired clusters in solution sets generated by evolutionary diversity optimization (EDO), which seeks to compute diverse solutions of high quality. We demonstrate that employing lp-norms when designing a diversity measure based on pairwise comparisons discourages clusters in a population, in accordance with the intuitive notion of diversity. Furthermore, we propose a novel diversity measure specifically tailored for parallel machine scheduling, leveraging direct sequential relationships between job pairs. Through experimental validation, we demonstrate that integrating our diversity measure into an established evolutionary algorithm yields highly diverse solution sets and show that the use of lp-norms leads to solution sets exhibiting higher robustness than established methods, enabling better adaptability to subsequent modifications of the model.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(xu:2025:GECCO, author = {Li-Ting Xu and Qiang Yang and Dan-Ting Duan and Xin Lin and Cheng-Zhi Qu and Zhen-Yu Lu and Jun Zhang}, title = {Ant Colony Optimization for Tourist Route Planning}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Sarah L. Thomson and Yi Mei}, pages = {349--357}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {ant colony optimization, tourist route planning, constrained optimization, combinational optimization, Evolutionary Combinatorial Optimization, Metaheuristics}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726341}, doi = {doi:10.1145/3712256.3726341}, size = {9 pages}, abstract = {This paper develops a new Tourist Route Planning (TRP) model by incorporating the entrance fees and the experience values of scenic spots, the travelling costs between scenic spots, and the budget of the tourist. Resultantly, the new TRP aims at finding an optimal route by maximizing the travelling experience value of the tourist with the constraint that the total cost of the route including the travelling costs and the spot entrance fees does not exceed the given budget. To effectively solve this new TRP, this paper adapts the five classical ant colony optimization algorithms (ACO), namely ant system (AS), elite AS (EAS), rank-based AS (RAS), max-min AS (MMAS), and ant colony system (ACS). To this end, this paper first introduces a new heuristic information measure by integrating the experience values and the entrance fees of the scenic spots, and the traveling costs between scenic spots. Further, a new local search strategy encompassing 2-opt and one spot insertion operator is designed to further improve the quality of the route under the budget constraint. Abundant experiments have been carried out on various TRP instances of three scales, namely small-scale, medium-scale, and large-scale, involving different numbers of scenic spots and different settings of budgets. The experimental results demonstrate that all the adapted five ACO algorithms are very effective for addressing the new TRP. Among them, RAS performs the best on small-scale TRP instances, and ACS obtains the best results on medium-scale TRP instances, while MMAS is the most effective one in addressing large-scale TRP instances.}, notes = {GECCO-2025 ECOM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(de-la-torre:2025:GECCO, author = {Camilo {De La Torre} and Giorgia Nadizar and Yuri Lavinas and Herve Luga and Dennis Wilson and Sylvain Cussat-Blanc}, title = {Evolution of Inherently Interpretable Visual Control Policies}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {358--367}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726332}, doi = {doi:10.1145/3712256.3726332}, size = {10 pages}, abstract = {Vision-based decision-making tasks encompass a wide range of applications, including safety-critical domains where trustworthiness is as key as performance. These tasks are often addressed using Deep Reinforcement Learning (DRL) techniques, based on Artificial Neural Networks (ANNs), to automate sequential decision making. However, the "black-box" nature of ANNs limits their applicability in these settings, where transparency and accountability are essential. To address this, various explanation methods have been proposed; however, they often fall short in fully elucidating the decision-making pipeline of ANNs, a critical aspect for ensuring reliability in safety-critical applications. To bridge this gap, we propose an approach based on Graph-based Genetic Programming (GGP) to generate transparent policies for vision-based control tasks. Our evolved policies are constrained in size and composed of simple and well-understood operational modules, enabling inherent interpretability. We evaluate our method on three Atari games, comparing explanations derived from common explainability techniques to those derived from interpreting the agent's true computational graph. We demonstrate that interpretable policies offer a more complete view of the decision process than explainability methods, enabling a full comprehension of competitive game-playing policies.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(gonzalez:2025:GECCO, author = {Everardo Gonzalez and Gaurav Dixit and Kagan Tumer}, title = {Dynamic Influence For Coevolutionary Agents}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {368--376}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726455}, doi = {doi:10.1145/3712256.3726455}, size = {9 pages}, abstract = {Multiagent settings are naturally characterized by coevolutionary dynamics, where agents must adapt and learn in the context of their teammates. A key challenge in such domains is determining how to credit an individual agent for their contribution to team performance. Fitness shaping approaches partially address this by identifying and isolating an agent's direct contribution to the team's success. However, when an agent's contribution is indirect-such as influencing other teammates to succeed-existing methods fail to account for its influence on the team. This paper introduces Dynamic Influence, a fitness shaping method for heterogeneous teams that isolates both direct and indirect contributions by evaluating how agents influence others over time. By considering inter-agent influence at a high temporal resolution, Dynamic Influence-Based Fitness Shaping allows agents to distill and extract direct credit from indirect interactions. Results in an autonomous aerial and terrestrial vehicle coordination problem demonstrate the efficacy of Dynamic Influence-Based Fitness Shaping, achieving superior cooperative behaviors compared to several static fitness shaping baselines.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(harn:2025:GECCO, author = {Po-Wei Harn and Bo Hui and Libo Sun and Wei-Shinn Ku}, title = {Evolutionary Quadtree Pooling for Convolutional Neural Networks}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {377--385}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726325}, doi = {doi:10.1145/3712256.3726325}, size = {9 pages}, abstract = {Despite the success of Convolutional Neural Networks (CNNs) in computer vision, it can be beneficial to reduce parameters, increase computational efficiency, and regulate overfitting. One such reduction technique is the use of so-called pooling, which gradually reduces the spatial dimensions of the data throughout the network. Recently, Quadtree-based Genetic Programming has achieved state-of-the-art results for optimizing spatial areas on customized requirements in different grid structures. Motivated by its success, we propose to extend this approach to pooling layers of CNNs. In this direction, this paper introduces a new way to look at each pooling layer. Specifically, we propose an Evolutionary Quadtree Pooling (EQP) method that can identify the best pooling scheme. By embedding multiple quadtrees set as a pooling scheme in the pooling layers of a CNN, we are able to operate crossover and mutation on the feature maps. The evolutionary process of EQP guides the search to provide more reliable evaluations, where each individual can be seen as a CNN with a new type of pooling scheme. Our experimental results show that the best candidate network of EQP outperforms state-of-the-art max, average, stochastic, median, soft, and mixed pooling in accuracy and overfitting reduction while maintaining low computational costs. Our codes are available at https://github.com/poweiharn/EQP.git.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(jorgensen:2025:GECCO, author = {Steven Jorgensen and Erik Hemberg and Jamal Toutouh and Una-May O'Reilly}, title = {Guiding Evolutionary {AutoEncoder} Training with Activation-Based Pruning Operators}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {386--396}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726449}, doi = {doi:10.1145/3712256.3726449}, size = {11 pages}, abstract = {This study explores a novel approach to neural network pruning using evolutionary computation, focusing on simultaneously pruning the encoder and decoder of an autoencoder. We introduce two new mutation operators that use layer activations to guide weight pruning. Our findings reveal that one of these activation-informed operators outperforms random pruning, resulting in more efficient autoencoders with comparable performance to canonically trained models. Prior work has established that autoencoder training is effective and scalable with a spatial coevolutionary algorithm that cooperatively coevolves a population of encoders with a population of decoders, rather than one autoencoder. We evaluate how the same activity-guided mutation operators transfer to this context. We find that random pruning is better than guided pruning, in the coevolutionary setting. This suggests activation-based guidance proves more effective in low-dimensional pruning environments, where constrained sample spaces can lead to deviations from true uniformity in randomization. Conversely, population-driven strategies enhance robustness by expanding the total pruning dimensionality, achieving statistically uniform randomness that better preserves system dynamics. We experiment with pruning according to different schedules and present best combinations of operator and schedule for the canonical and coevolving populations cases.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(le:2025:GECCO, author = {Khang Gia Le and Ngoc Hoang Luong}, title = {Black-Box Adversarial Attack on Dialogue Generation via Multi-Objective Optimization}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {397--406}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726431}, doi = {doi:10.1145/3712256.3726431}, size = {10 pages}, abstract = {Transformer-based dialogue generation (DG) models are ubiquitous in modern conversational artificial intelligence platforms. These models, however, are susceptible to adversarial attacks, i.e., prompts that appear indiscernible from normal inputs but are maliciously crafted to make the models generate incoherent and irrelevant responses. Evaluating the adversarial robustness of DG models is crucial to their real-world deployment. Adversarial methods typically exploit gradient information to effectively modify key input tokens, thereby achieving excellent attack performance. Nevertheless, such white-box approaches are impractical in real-world scenarios since the models' internal parameters are inaccessible. While black-box methods, which exploit only input prompts and DG models' output responses, offer a wider applicability, they often suffer from poor performance. In a human-machine conversation, good responses are expected to be semantically coherent and textually succinct. We formulate adversarial attack on DG models as a bi-objective optimization problem, where input prompts are modified in order to minimize the response coherence and maximize the generation length. We propose a DG attack framework (DGAttack) that employs multi-objective optimization to consider both objectives simultaneously when perturbing user prompts to craft adversarial inputs. Experiments across four benchmark datasets and (large) language models demonstrate the excellent performance of DGAttack compared to existing state-of-the-art approaches.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(lipschutz-villa:2025:GECCO, author = {Gabriel Lipschutz-Villa and Harsh Bandhey and Ruonan Yin and Malek Kamoun and Ryan Urbanowicz}, title = {Rule-based Machine Learning: Separating Rule and Rule-Set Pareto-Optimization for Interpretable Noise-Agnostic Modeling}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {407--415}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {rule-based machine learning, learning classifier systems, supervised learning, interpretability, multi-objective optimization, Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726461}, doi = {doi:10.1145/3712256.3726461}, size = {9 pages}, abstract = {Rule-based machine learning (RBML) algorithms, e.g. learning classifier systems (LCSs), can capture complex relationships while yielding more interpretable models than most other machine learning algorithms. Traditional LCSs rely on a single fitness function for both rule and/or rule-set optimization. However, ideal rule vs. rule-set discovery often requires distinct and multiple objectives. Recently, hybrid-LCSs were proposed that explicitly separated the task of rule vs. rule-set discovery but relied on distinct single-objective or weighted multi-objective fitness functions. This study introduces a newly developed Heuristic Evolutionary Rule Optimization System (HEROS) that combines previous LCS innovations aimed at tackling noisy, larger-scale, classification tasks, while adopting separation of rule vs. rule-set evolution. Uniquely, HEROS employs a custom Pareto-front-based multi-objective fitness function (for rule discovery) and NSGA-II-style multi-objective optimization (for rule-set discovery) to solve both clean and noisy-signal classification problems agnostically. Rule discovery is driven by rule-accuracy and instance coverage objectives, while rule-set discovery is driven by prediction accuracy and rule-set size objectives. Using diverse simulated benchmark datasets, i.e. noisy (GAMETES) and clean (MUX), we demonstrate proof-of-principle that HEROS can directly discover accurate, highly-compact, interpretable, and ideal solutions when compared to the established 'ExSTraCS' RBML algorithm, without objective weightings or adjusting hyperparameters.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(mateos-melero:2025:GECCO, author = {Enrique Mateos-Melero and Miguel {Iglesias Alcazar} and Raquel Fuentetaja and Fernando Fernandez}, title = {Dataset Reduction for Offline Reinforcement Learning using Genetic Algorithms with Image-Based Heuristics}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {416--424}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {offline reinforcement learning, genetic algorithms, dataset quality, learning performance prediction, convolutional neural networks, Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726364}, doi = {doi:10.1145/3712256.3726364}, size = {9 pages}, abstract = {In offline Reinforcement Learning (RL), the size and quality of the training dataset play a crucial role in determining policy performance. Large datasets can lead to excessive training times, while low-quality data can result in sub-optimal policies, particularly for deep learning-based RL frameworks. To address these challenges, we propose a novel approach that leverages genetic algorithms for efficient dataset reduction, paired with image-based learning using Convolutional Neural Networks (CNNs) to reduce the evaluation time of the fitness function. Specifically, our method predicts the performance of policies (fitness) learned from offline RL datasets (phenotype) and identifies optimized subsets that preserve or enhance policy quality. We evaluate our approach across three well-established RL domains, demonstrating that it effectively reduces dataset size while maintaining or improving policy performance. Furthermore, we show the transferability of the learned models to similar tasks, enabling efficient dataset optimization via transfer learning.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(merry:2025:GECCO, author = {Michael Merry and Patricia Riddle and James Warren}, title = {{PropNEAT} - Efficient {GPU-Compatible} Backpropagation over {NeuroEvolutionary} Augmenting Topology Networks}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {425--433}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726319}, doi = {doi:10.1145/3712256.3726319}, size = {9 pages}, abstract = {We introduce PropNEAT, a fast backpropagation implementation of NEAT that uses a bidirectional mapping of the genome graph to a layer-based architecture that preserves the NEAT genomes whilst enabling efficient GPU backpropagation. We test PropNEAT on 58 binary classification datasets from the Penn Machine Learning Benchmarks database, comparing the performance against logistic regression, dense neural networks and random forests, as well as a densely retrained variant of the final PropNEAT model. Random forests was the highest performer, with no significant difference between it, PropNEAT, PropNEAT-retrain or dense neural networks. Logistic regression was significantly worse than all other models. PropNEAT was faster than a naive NEAT backpropagation method, and both were faster and had better performance than the original NEAT implementation. The per-epoch training time for PropNEAT scales linearly with network depth, and is efficient on GPU implementations for backpropagation. This implementation could be extended to support reinforcement learning or convolutional networks, and is able to find well-performing, sparser and smaller networks with potential for applications in low-power contexts.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(nesmachnow:2025:GECCO, author = {Sergio Nesmachnow and Jamal Toutouh}, title = {Adversarial attacks to image classification systems using evolutionary algorithms}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {434--442}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726429}, doi = {doi:10.1145/3712256.3726429}, size = {9 pages}, abstract = {Image classification currently faces significant security challenges due to adversarial attacks, which consist of intentional alterations designed to deceive classification models based on artificial intelligence. This article explores an approach to generate adversarial attacks against image classifiers using a combination of evolutionary algorithms and generative adversarial networks. The proposed approach explores the latent space of a generative adversarial network with an evolutionary algorithm to find vectors representing adversarial attacks. The approach was evaluated in two case studies corresponding to the classification of handwritten digits and object images. The results showed success rates of up to 35\% for handwritten digits, and up to 75\% for object images, improving over other search methods and reported results in related works. The applied method proved to be effective in handling data diversity on the target datasets, even in problem instances that presented additional challenges due to the complexity and richness of information.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(pan:2025:GECCO, author = {Shuaiqun Pan and Yash J. Patel and Aneta Neumann and Frank Neumann and Thomas Baeck and Hao Wang}, title = {Evolving Hard Maximum Cut Instances for Quantum Approximate Optimization Algorithms}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {443--452}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {variational quantum algorithms, RQAOA, goemans and williamson algorithm, graph autoencoder, Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726415}, doi = {doi:10.1145/3712256.3726415}, size = {10 pages}, abstract = {Variational quantum algorithms, such as the Recursive Quantum Approximate Optimization Algorithm (RQAOA), have become increasingly popular, offering promising avenues for employing Noisy Intermediate-Scale Quantum devices to address challenging combinatorial optimization tasks like the maximum cut problem. In this study, we use an evolutionary algorithm equipped with a unique fitness function. This approach targets hard maximum cut instances within the latent space of a Graph Autoencoder, identifying those that pose significant challenges or are particularly tractable for RQAOA, in contrast to the classic Goemans and Williamson algorithm. Our findings not only delineate the distinct capabilities and limitations of each algorithm but also expand our understanding of RQAOA's operational limits. Furthermore, the diverse set of graphs we have generated serves as a crucial benchmarking asset, emphasizing the need for more advanced algorithms to tackle combinatorial optimization challenges. Additionally, our results pave the way for new avenues in graph generation research, offering exciting opportunities for future explorations.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(rovito:2025:GECCO, author = {Luigi Rovito and Marco Virgolin}, title = {Interpretable Non-linear Survival Analysis with Evolutionary Symbolic Regression}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {453--462}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726446}, doi = {doi:10.1145/3712256.3726446}, size = {10 pages}, abstract = {Survival Regression (SuR) is a key technique for modeling time to event in important applications such as clinical trials and semiconductor manufacturing. Currently, SuR algorithms belong to one of three classes: non-linear black-box-allowing adaptability to many datasets but offering limited interpretability (e.g., tree ensembles); linear glass-box-being easier to interpret but limited to modeling only linear interactions (e.g., Cox proportional hazards); and non-linear glass-box-allowing adaptability and interpretability, but empirically found to have several limitations (e.g., explainable boosting machines, survival trees). In this work, we investigate whether Symbolic Regression (SR), i.e., the automated search of mathematical expressions from data, can lead to non-linear glassbox survival models that are interpretable and accurate. We propose an evolutionary, multi-objective, and multi-expression implementation of SR adapted to SuR. Our empirical results on five real-world datasets show that SR consistently outperforms traditional glassbox methods for SuR in terms of accuracy per number of dimensions in the model, while exhibiting comparable accuracy with black-box methods. Furthermore, we offer qualitative examples to assess the interpretability potential of SR models for SuR. Code at: https://github.com/lurovi/SurvivalMultiTree-pyNSGP.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(sobania:2025:GECCO, author = {Dominik Sobania and Martin Briesch and Franz Rothlauf}, title = {ImageBreeder: Guiding Diffusion Models with Evolutionary Computation}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {463--471}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726439}, doi = {doi:10.1145/3712256.3726439}, size = {9 pages}, abstract = {With the recent advancements of diffusion models, it is quite easy to generate high-quality images. However, many attempts and manual changes are often necessary to achieve this high quality. Leveraging evolutionary algorithms to automate this process therefore presents a promising approach. Consequently, we introduce Image-Breeder as a framework to improve image generation at inference time driven by evolutionary algorithms. Additionally, we study the effectiveness of 10 different variation operators ranging from pixel-based blending techniques to modifications directly on the latent representation of the images. The results show that using evolutionary algorithms can significantly increase an image's quality as well as the alignment with a given prompt. Furthermore, all tested guided search methods are human competitive, as a random trial and error approach is outperformed on over 75\% of the benchmark problems. In addition to the empirical results, we also critically discuss the benefits and challenges of our framework uncovering future research directions. We recommend researchers to focus on optimising latent image representations for this purpose as this may improve the ability of evolutionary algorithms to transfer promising features from one image to another, unlocking even better image generation.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(teixeira:2025:GECCO, author = {Matheus Candido Teixeira and Gisele Lobo Pappa}, title = {Transformers as Surrogate Models for Genetic Programming in {AutoML} Tasks}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {472--480}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726396}, doi = {doi:10.1145/3712256.3726396}, size = {9 pages}, abstract = {In applications where the fitness function has a high computational cost, one of the main drawbacks of Evolutionary Algorithms when compared to other search methods is a prohibitive computational cost. The use of surrogates as proxies for fitness function calculation to alleviate this problem is not new, but addressing the problem as a binary relation learning, i.e., evaluating if one individual is better or worse than another without estimating the actual value of the fitness, is a recent trend.This paper proposes a transformer-encoder as a surrogate to evaluate pairs of solutions and determine their relationship, i.e., which one is better/worse than the other. We experimented the model in the context of AutoML, which seeks to find the best combination of algorithms for a classification problem. To optimize the pipeline, we can use a genetic programming, but the cost of evaluating each individual is generally expensive.We trained the encoder with several parameters and compared its performance against traditional GP - evaluating fitness at each generation. Results confirm using the encoder as a surrogate does not degrade the fitness values of the evolved population of ML pipelines and can even improve it in some cases (up to 285 times faster).}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(torrijos:2025:GECCO, author = {Pablo Torrijos and Jose A. Gamez and Jose M. Puerta and Juan A. Aledo}, title = {Genetic Algorithms for Tractable Bayesian Network Fusion via Pre-Fusion Edge Pruning}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {481--489}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726333}, doi = {doi:10.1145/3712256.3726333}, size = {9 pages}, abstract = {Bayesian Network (BN) fusion combines multiple input networks into a single structure, balancing dependency preservation with computational tractability. While unrestricted fusion retains all dependencies, it often results in overly complex networks with high treewidth, which affects inference scalability. Limited fusion mitigates this by pruning edges to control treewidth but risks overfitting to input-specific noise and omitting dependencies from the original BNs. This paper introduces a consensus framework that prioritizes shared structures among input networks while enforcing treewidth constraints, ensuring a good consensus. We propose genetic algorithms with advanced initialization, specialized operators, and a tailored fitness function. Additionally, we adapt existing methods to this problem and implement greedy baselines for benchmarking and further optimization. Experiments on synthetic and real-world BNs show the superiority of the proposed genetic algorithms over the adapted methods and greedy baselines.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(uchikoshi:2025:GECCO, author = {Motonobu Uchikoshi and Youhei Akimoto}, title = {Feature selection based on cluster assumption in {PU} learning}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {490--498}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {feature selection, positive unlabeled learning, cluster assumption, Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726345}, doi = {doi:10.1145/3712256.3726345}, size = {9 pages}, abstract = {Feature selection is essential for efficient data mining and sometimes encounters the positive-unlabeled (PU) learning scenario, where only a few positive labels are available, while most data remains unlabeled. In certain real-world PU learning tasks, data subjected to adequate feature selection often form clusters with concentrated positive labels. Conventional feature selection methods that treat unlabeled data as negative may fail to capture the statistical characteristics of positive data in such scenarios, leading to suboptimal performance. To address this, we propose a novel feature selection method based on the cluster assumption in PU learning, called FSCPU. FSCPU formulates the feature selection problem as a binary optimization task, with an objective function explicitly designed to incorporate the cluster assumption in the PU learning setting. Experiments on synthetic datasets demonstrate the effectiveness of FSCPU across various data conditions. Moreover, comparisons with 10 conventional algorithms on three open datasets show that FSCPU achieves competitive performance in downstream classification tasks, even when the cluster assumption does not strictly hold.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(uwano:2025:GECCO, author = {Fumito Uwano and Will N. Browne}, title = {Enhancing {XCS} with Dual-Stream Identification for Perceptual Aliasing in Multi-Step Decision-Making}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {499--507}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726470}, doi = {doi:10.1145/3712256.3726470}, size = {9 pages}, abstract = {Perceptual aliasing, where distinct states appear indistinguishable due to sensor limitations or environmental ambiguities, poses significant challenges in multi-step decision-making. The eXtended Classifier System (XCS) addresses this issue by identifying unique state transition patterns and combining them to construct accurate policies. Additionally, state-action-state chains enhance XCS's ability to handle sequentially aliased states. However, XCS processes aliased states sequentially as they are perceived, which can lead to performance degradation when incorrect versions of aliased states are included in the chain. This limitation underscores the need for a more robust mechanism to accurately differentiate unique states from aliased ones to ensure reliable policy creation. To address this, we propose a dual-stream identification framework that enhances XCS's performance in environments with perceptual aliasing. The framework introduces two parallel identification processes: one captures immediate state-action relationships, while the other identifies broader patterns across multi-step sequences. By integrating these dual streams, the proposed approach effectively disambiguates aliased states, enabling more accurate decision-making. Experimental evaluations demonstrate that our dual-stream model outperforms state-of-the-art XCS implementations across 14 benchmark environments.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(yadav:2025:GECCO, author = {Deepanshu Yadav and Palaniappan Ramu and Kalyanmoy Deb}, title = {Machine Learning-Assisted Constraint Handling Under Variable Uncertainty for Preference-based Multi-Objective Optimization}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {508--516}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {evolutionary algorithms, multi-criteria decision-making, machine learning, reference point, reliability, uncertainty, Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726462}, doi = {doi:10.1145/3712256.3726462}, size = {9 pages}, abstract = {Evolutionary Multi-objective Optimization (EMO) algorithms are widely used to solve real-world multi-objective optimization problems, aiming to obtain a set of non-dominated solutions close to the Pareto front. However, most EMO methods assume deterministic decision variables, ignoring inherent uncertainties in engineering applications, which can lead to design failures, especially in reliability-based designs. Reliability-based Multi-objective Optimization (ReMOO) addresses this issue by incorporating variable uncertainty and probabilistic constraints to generate a Reliable Front. ReMOO operates using a bi-level framework: the outer level optimizes objective functions, while the inner level estimates reliability through computationally intensive methods, like Monte Carlo Simulation (MCS) or the Performance Measure Approach (PMA). Additionally, decision-makers (DMs) often select only a subset of reliable solutions, limiting computational efficiency. To overcome these challenges, this paper proposes a Machine Learning-assisted reliability-based Multi-Criteria Decision-Making (ML-ReMCDM) technique. ML models are trained on reliability-based constraints within the decision space before an EMO execution. In the inner loop, ML models predict probabilistic constraints and reliability indices, significantly reducing computational costs. Moreover, the outer loop computes only the DM-preferred segment of the reliable front, further enhancing efficiency. The ML-ReMCDM approach, implemented on several benchmark and real-world examples, demonstrates substantial improvements in computational efficiency as well as practical applicability.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(zhang:2025:GECCO, author = {Yisong Zhang and Guoxing Yi}, title = {{LAOS:} Large Language Model-Driven Adaptive Operator Selection for Evolutionary Algorithms}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Ryan Urbanowicz and Will N. Browne}, pages = {517--526}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Machine Learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726450}, doi = {doi:10.1145/3712256.3726450}, size = {10 pages}, abstract = {Adaptive Operator Selection (AOS) is a strategy in Evolutionary Algorithms (EAs) that dynamically adjusts the application frequency of operators to enhance search efficiency based on online performance feedback. This paper introduces LAOS, an AOS framework driven by Large Language Models (LLMs). We design a meta-prompt to provide optimization state information (such as optimization progress, best fitness, and population diversity) and operator credit assignment, assisting LLMs in making adaptive decisions. Furthermore, LAOS maintains a dual-layer replay buffer structure: the offline layer records historical experiences under fixed operator strategies, while the online layer accumulates dynamically generated experiences during execution. By employing a similar experience sampling strategy, the framework can provide decision-making support for LLMs, enhancing both the efficiency and accuracy of search strategies. Experimental results on continuous numerical optimization and three categories of combinatorial optimization problems validate the effectiveness and generalization capability of LAOS. This study demonstrates the feasibility of leveraging LLMs for AOS, showcasing their potential in enhancing optimization performance and supporting automated algorithm design.}, notes = {GECCO-2025 EML A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(auger:2025:GECCO, author = {Anne Auger and Dimo Brockhoff and Jordan Cork and Tea Tusar}, title = {On the Pareto Set and Front of Multiobjective Spherical Functions with Convex Constraints}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {527--535}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726432}, doi = {doi:10.1145/3712256.3726432}, size = {9 pages}, abstract = {We analyze a fundamental class of multiobjective constrained problems where the objectives are spherical functions and the constraints are convex. As an application from the projection theorem on closed convex sets, we prove that the constrained Pareto set corresponds to the orthogonal projection of the unconstrained Pareto set onto the feasible region. We establish this fundamental geometric property and illustrate its implications using visualizations of Pareto sets and fronts under various constraint configurations. Furthermore, we assess the performance of NSGA-II on these problems, examining its ability to approximate the constrained Pareto set across different dimensions. Our findings highlight the importance of theoretically grounded and understood benchmark problems for assessing algorithmic behavior and contribute to a deeper understanding of constrained multiobjective landscapes.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(bostelmann-arp:2025:GECCO, author = {Lukas Bostelmann-Arp and Christoph Steup and Sanaz Mostaghim}, title = {Genotype vs. Phenotype: A Crossover Operator Comparison for the Multi-Objective Coverage Path Planning Problem}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {536--544}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {evolutionary multi-objective optimization, crossover operator, continuous coverage path planning, Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726350}, doi = {doi:10.1145/3712256.3726350}, size = {9 pages}, abstract = {The crossover operator is a fundamental component of genetic algorithms, combining genetic material from parent solutions to generate offspring. Traditionally, crossover is performed in the search space using the genotype. However, it can also be executed in the solution space on the phenotype, offering potential advantages such as improved feasibility preservation, faster convergence, and greater explainability. These benefits, however, come with tradeoffs, including increased implementation complexity, higher computational costs, and a likely reduction in solution diversity. This study examines the properties of search space and solution space crossover operators in the context of a multi-objective, weighted, and continuous coverage path planning problem. Three crossover strategies are tested: two of which operate directly on the genotype and one that uses intersections of the phenotype.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(chen:2025:GECCO, author = {Yanyu Chen and Hisao Ishibuchi and Yang Nan}, title = {Influence of Subpopulation on the Performance of Coevolutionary Algorithms for Constrained Multiobjective Optimization Problems}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {545--553}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {evolutionary multiobjective optimization (EMO), coevolutionary algorithms, constraint handling technique (CHT), Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726344}, doi = {doi:10.1145/3712256.3726344}, size = {9 pages}, abstract = {To effectively use the subpopulation in a coevolutionary algorithm for constrained multiobjective optimization problems (CMOPs), this paper divides its search process into two stages with different search priorities, leveraging the stage-switching mechanism of PPS-MOEA/D. Based on the CCMO framework, we propose several variants with different subpopulation strategies in the second stage, and evaluate their performance on both artificial and real-world CMOPs. Our experimental results reveal that CMOPs can be classified into three categories. The results also confirm the usefulness of the two strategies in the second stage: (i) to decrease the number of offspring generated in the subpopulation, and (ii) to enhance the cooperation (connection) between the main population and the subpopulation.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(glasmachers:2025:GECCO, author = {Tobias Glasmachers}, title = {Variable Metric Evolution Strategies for High-dimensional Multi-Objective Optimization}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {554--562}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726304}, doi = {doi:10.1145/3712256.3726304}, size = {9 pages}, abstract = {We design a class of variable metric evolution strategies well suited for high-dimensional problems. We target problems with many variables, not (necessarily) with many objectives. The construction combines two independent developments: efficient algorithms for scaling covariance matrix adaptation to high dimensions, and evolution strategies for multi-objective optimization. In order to design a specific instance of the class we first develop a (1+1) version of the limited memory matrix adaptation evolution strategy and then use an established standard construction to turn a population thereof into a state-of-the-art multi-objective optimizer with indicator-based selection. The method compares favorably to adaptation of the full covariance matrix.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(honjo-ide:2025:GECCO, author = {Felipe {Honjo Ide} and Hernan Aguirre and Kiyoshi Tanaka}, title = {An Evolutionary Algorithm for Solving Decision Space Constrained Multi-Objective Binary Optimization Problems}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {563--571}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, combinatorial optimization, multi-objective optimization, constraint handling, Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726374}, doi = {doi:10.1145/3712256.3726374}, size = {9 pages}, abstract = {In real-world multi-objective optimization problems, it is common to find constraints that limit the feasible space, challenging the solver to explore the infeasible region and find good feasible solutions. Several evolutionary algorithms with various constraint-handling techniques have been proposed over the years. However, most focus on problems with continuous variables and constraints defined over the objective space and might not be suitable for binary problems and constraints defined on the decision space. This work proposes a multi-objective evolutionary algorithm for solving decision space-constrained multi-objective binary optimization problems. The proposed method can switch between a simple evolutionary algorithm, which optimizes constraint violation of infeasible solutions, and a random bit climber, which optimizes the objective functions of feasible solutions. We compare the performance of the proposed algorithm to other state-of-the-art evolutionary algorithms and study its behavior using SAT Constrained MNK-Landscapes. We show that the proposed algorithm can effectively optimize constraint violation of infeasible solutions, quickly find feasible solutions, and performs better than the compared algorithms in highly constrained problems with varying numbers of objectives, epistatic interactions, equality and inequality constraints, and constraint difficulty.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(horaguchi:2025:GECCO, author = {Yuma Horaguchi and Masaya Nakata}, title = {High-Dimensional Expensive Multiobjective Optimization Using a Surrogate-Assisted Multifactorial Evolutionary Algorithm}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {572--580}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {expensive multiobjective optimization, surrogate-assisted evolutionary algorithm, high-dimensional, multifactorial evolutionary algorithm, Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726483}, doi = {doi:10.1145/3712256.3726483}, size = {9 pages}, abstract = {The performance of surrogate-assisted multiobjective evolutionary algorithms (SAMOEAs) often degrades in high-dimensional problems. Recent studies have shown that decomposition-based approaches are particularly effective in handling high-dimensional search spaces, owing to their problem-simplifying capability. However, existing decomposition-based SAMOEAs are designed to sequentially solve each decomposed subproblem, still unnecessarily consuming function evaluations (FEs) and thus degrading the search efficiency. To address this issue, this paper proposes a novel decomposition-based SAMOEA that employs a multifactorial evolutionary algorithm (MFEA). The proposed algorithm aggregates multiple subproblems randomly and it collectively solves them using a surrogate-assisted MFEA framework. This approach enables the efficient discovery of promising solutions across multiple subproblems in a single FE, enhancing the search efficiency under a limited budget of FEs. Experimental results show that our proposed algorithm outperforms state-of-the-art SAMOEAs on problems with up to 300 dimensions. This suggests that our surrogate-assisted MFEA framework can bring out the further potential of decomposition-based SAMOEAs.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(ishibuchi:2025:GECCO, author = {Hisao Ishibuchi and Lie Meng Pang and Cheng Gong}, title = {Search Behavior Analysis of {NSGA-III:} Dominance-based and Decomposition-based Multi-objective Evolutionary Algorithm}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {581--589}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726433}, doi = {doi:10.1145/3712256.3726433}, size = {9 pages}, abstract = {In the field of evolutionary multi-objective optimization (EMO), EMO algorithms are often categorized into three types: dominance-based, decomposition-based and indicator-based algorithms. In this categorization, NSGA-III is handled as a decomposition-based algorithm. This is because a single solution is assigned to each of uniformly generated reference vectors as in MOEA/D. However, in a recent survey paper, NSGA-III was categorized in the same group as NSGA-II based on their generation update mechanisms. Another recent study demonstrated that NSGA-III shows a similar search behavior to NSGA-II for combinatorial multi-objective problems. However, for DTLZ test problems, NSGA-III shows almost the same search behavior as MOEA/D. In this paper, we demonstrate that the shape of the Pareto front is the main factor about the search behavior of NSGA-III. If a test problem has a regular (i.e., triangular) Pareto front, NSGA-III shows the same search behavior as MOEA/D. However, if a test problem has an irregular Pareto front (e.g., inverted triangular), NSGA-III shows a similar search behavior to NSGA-II. We also demonstrate that the objective space normalization in NSGA-III is not stable for multi-objective problems with inverted triangular Pareto fronts.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(korogi:2025:GECCO, author = {Keisuke Korogi and Ryoji Tanabe}, title = {Analyzing the Landscape of the Indicator-based Subset Selection Problem}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {590--599}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726381}, doi = {doi:10.1145/3712256.3726381}, size = {10 pages}, abstract = {The indicator-based subset selection problem (ISSP) involves finding a point subset that minimizes or maximizes a quality indicator. The ISSP is frequently found in evolutionary multi-objective optimization (EMO). An in-depth understanding of the landscape of the ISSP could be helpful in developing efficient subset selection methods and explaining their performance. However, the landscape of the ISSP is poorly understood. To address this issue, this paper analyzes the landscape of the ISSP by using various traditional landscape analysis measures and exact local optima networks (LONs). This paper mainly investigates how the landscape of the ISSP is influenced by the choice of a quality indicator and the shape of the Pareto front. Our findings provide insightful information about the ISSP. For example, high neutrality and many local optima are observed in the results for ISSP instances with the additive epsilon-indicator.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(larraga:2025:GECCO, author = {Giomara Larraga and Kaisa Miettinen}, title = {Exploring Phase-Specific Configuration of Interactive Evolutionary Multiobjective Optimization Methods}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {600--608}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {multiobjective optimization, decision making, interactive methods, evolutionary algorithms, Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726367}, doi = {doi:10.1145/3712256.3726367}, size = {9 pages}, abstract = {Interactive evolutionary multiobjective methods enable a decision maker to solve optimization problems involving multiple conflicting objective functions by iteratively incorporating preference information. When applying interactive methods, two phases can often be identified: a learning phase, where a decision maker gains insights on trade-offs and identifies a region of interest based on their preferences, and a decision phase focused on fine-tuning and selecting the most preferred solution. The configuration of evolutionary operators, like selection, crossover, and mutation, heavily influences the performance of evolutionary methods. However, despite extensive research on parameter tuning, identifying optimal configurations for these operators within interactive methods while accounting for the specific goals of each phase has not been studied. This study introduces a framework for the automatic configuration of interactive methods, taking the first step toward addressing this research gap. The framework systematically identifies phase-specific optimal configurations by combining the PHI indicator with the irace automatic configuration tool. Experiments with interactive RVEA and interactive RNSGA-II on problems involving three, five, and seven objective functions reveal notable differences in optimal configurations between the learning and decision phases. These findings lay a foundation for enhancing the performance of interactive evolutionary multiobjective methods and highlight the importance of phase-specific configurations.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(li:2025:GECCO, author = {Zhipan Li and Wenkai Mao and Huigui Rong and Jianguo Chen and Shengxu Huo and Zilu Zhao}, title = {Improved Convergence-relaxed Mechanism for Handling Imbalance Between Convergence and Diversity in the Decision Space in Multimodal Multi-objective optimization}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {609--617}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {multimodal multi-objective optimization, enhanced local convergence indicator, evolutionary algorithm, convergence degradation, Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726356}, doi = {doi:10.1145/3712256.3726356}, size = {9 pages}, abstract = {Balancing convergence and diversity in the decision space is essential in solving multimodal multi-objective optimization problems (MMOPs), which have multiple equivalent Pareto optimal sets (PSs) with the same Pareto optimal front (PF). For MMOPs with an imbalance between convergence and diversity in the decision space (MMOP-ICD), numerous efficient multimodal multiobjective evolutionary algorithms (MMEAs) avoid premature convergence and search for the imbalanced PS by relaxing the traditional convergence-first selection mechanism. Unfortunately, existing MMEAs suffer from convergence degradation due to excessive relaxation of the convergence-first selection mechanism. Therefore, this paper proposes an improved convergence-relaxed mechanism that includes an enhanced local convergence indicator and a two-stage mating selection. The enhanced local convergence indicator introduces the global convergence indicator into the local convergence indicator. The local convergence indicator can locate more equivalent PSs and prevent premature convergence caused by the global convergence indicator. The global convergence indicator can improve the convergence quality of the solution selected by the local convergence indicator. Then, the two-stage mating selection is used to enhance the diversity in the decision space and balance the improved convergence. Experimental results and statistical analysis show that the proposed algorithm is significantly superior to other state-of-the-art MMEAs.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(morales-paredes:2025:GECCO, author = {Adrian Isai Morales-Paredes and Jesus Guillermo Falcon-Cardona and Julio Juarez and Hugo Terashima-Marin and Carlos A. {Coello Coello}}, title = {Reference Point Specification in Greedy Inclusion Hypervolume-based Subset Selection: A Study on Two Objectives}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {618--626}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726438}, doi = {doi:10.1145/3712256.3726438}, size = {9 pages}, abstract = {The hypervolume indicator (HV) is widely used in evolutionary multi-objective optimization for performance evaluation and algorithm design. However, its utility heavily depends on selecting a reference point (RP). Depending on this point, HV may prefer non-uniform Pareto front approximations (PFAs) over other more uniform distributions. While existing studies have explored the impact of RP specification regarding mu-distribution settings, its impact on greedy inclusion hypervolume-based subset selection (GI-HSS) algorithms remains underexamined. These algorithms rely on incremental individual contributions to HV. This paper investigates the effect of the RP specification on the uniformity of PFAs generated by a GI-HSS algorithm. The focus is on two-objective Pareto fronts characterized by linear, concave, convex, and disconnected geometries. Using the lazy GI-HSS algorithm as a framework, we evaluate a comprehensive range of RP settings to identify those that promote more uniform distributions with up to 210 points. Our findings provide new insights into RP selection and offer practical guidelines for enhancing the performance of GI-HSS algorithms in multi-objective applications.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(nan:2025:GECCO, author = {Yang Nan and Hisao Ishibuchi and Tianye Shu and Ke Shang}, title = {R2 Indicator Analysis using the Optimal Distributions of Solutions for R2 and Other Indicators}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {627--635}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {evolutionary multi-objective optimization (EMO), R2 indicator, optimal distribution, Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726413}, doi = {doi:10.1145/3712256.3726413}, size = {9 pages}, abstract = {In the evolutionary multi-objective optimization (EMO) community, performance indicators are increasingly important. This is because the indicators can be used for evaluating and designing EMO algorithms. Among them, the hypervolume indicator is particularly popular because it is Pareto compliant. Recently, researchers have shown that the exact R2 indicator is also Pareto compliant. Some researchers have already investigated the optimal distribution of solutions for the hypervolume indicator. However, only a few studies have examined the optimal distribution of the approximate R2 indicator in the two-dimensional case. In this paper, we show the optimal distributions of solutions for the exact R2 indicator and some variants of the approximate R2 indicator in the three-dimensional case. The visualized optimal distributions are compared with each other and also with the best solution set for MOEA/D. Our analysis and results show that the optimal distribution of the exact R2 indicator is similar to that of the hypervolume indicator. The optimal distributions of the approximate R2 indicator's variants are similar to the best solution set for MOEA/D.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(nikolikj:2025:GECCO, author = {Ana Nikolikj and Gabriela Ochoa and Tome Eftimov}, title = {Customized Exploration of Landscape Features Driving Multi-Objective Combinatorial Optimization Performance}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {636--644}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726334}, doi = {doi:10.1145/3712256.3726334}, size = {9 pages}, abstract = {We present an analysis of landscape features for predicting the performance of multi-objective combinatorial optimization algorithms. We consider features from the recently proposed compressed Pareto Local Optimal Solutions Networks (C-PLOS-net) model of combinatorial landscapes. The benchmark instances are a set of rhomnk-landscapes with 2 and 3 objectives and various levels of ruggedness and objective correlation. We consider the performance of three algorithms - Pareto Local Search (PLS), Global Simple EMO Optimizer (GSEMO), and Non-dominated Sorting Genetic Algorithm (NSGA-II) - using the resolution and hypervolume metrics. Our tailored analysis reveals feature combinations that influence algorithm performance specific to certain landscapes. This study provides deeper insights into feature importance, tailored to specific rhomnk-landscapes and algorithms.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(santoshkumar:2025:GECCO, author = {Balija Santoshkumar and Kalyanmoy Deb}, title = {Addressing Heterogeneous Evaluation Times in Constrained Multi-Objective Optimization using a Mixed-Fidelity Evaluation Technique: Proof-of-Concept Results}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {645--653}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {constrained surrogate-assisted optimization, heterogeneous evaluation, multi-objective optimization, Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726362}, doi = {doi:10.1145/3712256.3726362}, size = {9 pages}, abstract = {Most practical optimization problems involve expensive evaluation procedures for computing objective and constraint functions. To obtain reasonable and accurate solutions close to true Pareto-optimal solutions, evolutionary multi-objective optimization (EMO) algorithms create surrogate models from already-evaluated high-fidelity solutions and use them during optimization to save computational time. However, most surrogate-assisted EMO algorithms are designed to evaluate all objectives and constraints of a solution, if found worthy of a high-fidelity evaluation. Such algorithms are inefficient if objectives and constraints involve heterogeneity with orders of magnitude of difference in evaluation times. Clearly, functions with relatively small evaluation time can be high-fidelity evaluated more often to obtain an overall idea of the potential importance of the solution before deciding to spend more time on evaluating expensive functions. In this paper, we propose an EMO approach that carefully determines which constraints and objectives should be high-fidelity evaluated for every population member and suggests a mixed-fidelity survival selection procedure capable of working with low- and high-fidelity evaluated population members. Results on a number of test and engineering problems indicate the viability of such a constrained multi- and many-objective optimization algorithm and encourage further attention.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(tanaka:2025:GECCO, author = {Shuhei Tanaka and Shoichiro Tanaka and Toshiharu Hatanaka}, title = {Scalarization-based Exploratory Landscape Analysis for Multi-Objective Continuous Optimization Problems}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {654--662}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {multi-objective optimization, exploratory landscape analysis, algorithm selection, feature-based method, Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726378}, doi = {doi:10.1145/3712256.3726378}, size = {9 pages}, abstract = {In landscape-aware techniques, interpretable features provide decision makers with insights into the relationship between algorithm behavior and problem properties. We focus on Exploratory Landscape Analysis (ELA), which is a well-established method for extracting human-designed features at low computational cost. Although the effectiveness of ELA has been proven in single-objective continuous optimization, extending it to multi-objective domains while maintaining both feature interpretability and effectiveness remains an open challenge. To address this challenge, we introduce Scalarization-based ELA (S-ELA), a novel variation of ELA for multi-objective continuous optimization using scalarizing methods. S-ELA enables the computation of conventional ELA features by scalarizing objective vectors. In this study, we investigated two scalarizing approaches: (1) decomposition and (2) non-dominated sorting. Through experiments on bi-objective continuous optimization problems from the bbob-biobj test suite, we compared S-ELA with Deep-ELA, a state-of-the-art deep learning-based ELA. Our results demonstrate that S-ELA achieved accuracies of approximately 76\% to 80\%, comparable to Deep-ELA, in algorithm selection.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(thakar:2025:GECCO, author = {Raghav Thakar and Gaurav Dixit and Siddarth Iyer and Kagan Tumer}, title = {Multiagent Credit Assignment for Multi-Objective Coordination}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {663--672}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726445}, doi = {doi:10.1145/3712256.3726445}, size = {10 pages}, abstract = {Many real-world coordination tasks-such as environmental monitoring, traffic management, and underwater exploration-are best modelled as multiagent problems with multiple, often conflicting objectives. Achieving effective coordination in these settings requires addressing two main challenges: 1) balancing multiple objectives and 2) resolving the credit assignment problem to isolate each agent's contribution from team-level feedback. Existing multiagent credit assignment methods collapse multi-objective reward vectors into a single scalar-potentially overlooking nuanced trade-offs. In this paper, we introduce the Multi-Objective Difference Evaluation (DMO) operator to assign agent-level credit without a priori scalarisation. DMO measures the change in hypervolume when an agent's policy is replaced by a counterfactual default, capturing how much that policy contributes to each objective and to the Pareto front. We embed DMO into the popular NSGA-II algorithm to evolve a population of joint policies with distinct trade-offs. Empirical results on the Multi-Objective Beach Problem and the Multi-Objective Rover Exploration domain show that our approach matches or surpasses existing baselines, delivering up to a 33\% performance improvement.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(zhao:2025:GECCO, author = {Shihan Zhao and Stefanos Nikolaidis}, title = {Multi-Objective Covariance Matrix Adaptation {MAP-Annealing}}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {673--682}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {quality diversity, multi-objective optimization, hypervolume, Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726463}, doi = {doi:10.1145/3712256.3726463}, size = {10 pages}, abstract = {Quality-Diversity (QD) optimization is an emerging field that focuses on finding a set of behaviorally diverse and high-quality solutions. While the quality is typically defined w.r.t. a single objective function, recent work on Multi-Objective Quality-Diversity (MOQD) extends QD optimization to simultaneously optimize multiple objective functions. This opens up multi-objective applications for QD, such as generating a diverse set of game maps that maximize difficulty, realism, or other properties. Existing MOQD algorithms use non-adaptive methods such as mutation and crossover to search for non-dominated solutions and construct an archive of Pareto Sets (PS). However, recent work in QD has demonstrated enhanced performance through the use of covariance-based evolution strategies for adaptive solution search. We propose bringing this insight into the MOQD problem, and introduce MO-CMA-MAE, a new MOQD algorithm that leverages Covariance Matrix AdaptationEvolution Strategies (CMA-ES) to optimize the hypervolume associated with every PS within the archive. We test MO-CMA-MAE on three MOQD domains, and for generating maps of a co-operative video game, showing significant improvements in performance.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(zuo:2025:GECCO, author = {Mingcheng Zuo and Dunwei Gong and Tianyang Xue and Chunliang Zhao and Yongde Guo}, title = {Constrained Multi-objective Optimization with Search Direction Learning}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Arnaud Liefooghe and Tapabrata Ray}, pages = {683--691}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {constrained multi-objective optimization, search direction learning, analog integrated circuit, Evolutionary Multiobjective Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726409}, doi = {doi:10.1145/3712256.3726409}, size = {9 pages}, abstract = {The solving process of constrained multi-objective evolutionary optimization algorithms (CMOEAs) is closely related to the search direction of the population. How to learn promising search directions through population data remains challenging. Therefore, this paper proposes a CMOEA with search direction learning. In this method, principal component analysis (PCA) is first used to learn the mainstream direction of population evolution, then single-constraint domination is used to learn the tributary direction of population evolution, and finally, the search directions are summarized to guide the generation of high-quality offspring populations. The performance comparison with five state-of-the-art algorithms on three standard test problems demonstrates the superiority of the proposed method. Its applicability in the field of simulated integrated circuits proves the scalability of the proposed method.}, notes = {GECCO-2025 EMO A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(akimoto:2025:GECCO, author = {Youhei Akimoto and Xilin Gao and Ze Kai Ng and Daiki Morinaga}, title = {Challenges of Interaction in Optimizing Mixed Categorical-Continuous Variables}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Tobias Glasmachers and Youhei Akimoto}, pages = {692--700}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {mixed binary-continuous optimization, evolution strategies, Evolutionary Numerical Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726370}, doi = {doi:10.1145/3712256.3726370}, size = {9 pages}, abstract = {Optimization of mixed categorical-continuous variables is prevalent in real-world applications of black-box optimization. Recently, CatCMA has been proposed as a method for optimizing such variables and has demonstrated success in hyper-parameter optimization problems. However, it encounters challenges when optimizing categorical variables in the presence of interaction between continuous and categorical variables in the objective function. In this paper, we focus on optimizing mixed binary-continuous variables as a special case and identify two types of variable interactions that make the problem particularly challenging for CatCMA. To address these difficulties, we propose two algorithmic components: a warm-starting strategy and a hyper-representation technique. We analyze their theoretical impact on test problems exhibiting these interaction properties. Empirical results demonstrate that the proposed components effectively address the identified challenges, and CatCMA enhanced with these components, named ICatCMA, outperforms the original CatCMA.}, notes = {GECCO-2025 ENUM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(de-nobel:2025:GECCO, author = {Jacob {de Nobel} and Diederick Vermetten and Hao Wang and Anna Kononova and Guenter Rudolph and Thomas Baeck}, title = {Abnormal Mutations: Evolution Strategies Don't Require Gaussianity}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Tobias Glasmachers and Youhei Akimoto}, pages = {701--709}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {evolution strategies, mutation distributions, gaussianity, benchmarking, CMA-ES, Evolutionary Numerical Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726479}, doi = {doi:10.1145/3712256.3726479}, size = {9 pages}, abstract = {The mutation process in evolution strategies has been interlinked with the normal distribution since its inception. Many lines of reasoning have been given for this strong dependency, ranging from maximum entropy arguments to the need for isotropy. However, some theoretical results suggest that other distributions might lead to similar local convergence properties. This paper empirically shows that a wide range of evolutionary strategies, from the (1+1)-ES to CMA-ES, show comparable optimization performance when using a mutation distribution other than the standard Gaussian. Replacing it with, e.g., uniformly distributed mutations, does not deteriorate the performance of ES, when using the default adaptation mechanism for the strategy parameters. We observe that these results hold not only for the sphere model but also for a wider range of benchmark problems.}, notes = {GECCO-2025 ENUM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(dinu:2025:GECCO, author = {Catalin-Viorel Dinu and Yash J. Patel and Xavier Bonet-Monroig and Hao Wang}, title = {An Adaptive Re-evaluation Method for Evolution Strategy under Additive Noise}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Tobias Glasmachers and Youhei Akimoto}, pages = {710--718}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {evolutionary strategy, lipschitz constant, noisy optimization, CMA-ES, black-box optimization, additive gaussian noise, Evolutionary Numerical Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726352}, doi = {doi:10.1145/3712256.3726352}, size = {9 pages}, abstract = {The Covariance Matrix Adaptation Evolutionary Strategy (CMA-ES) is one of the most advanced algorithms in numerical black-box optimization. For noisy objective functions, several approaches were proposed to mitigate the noise, e.g., re-evaluations of the same solution or adapting the population size. In this paper, we devise a novel method to adaptively choose the optimal re-evaluation number for function values corrupted by additive Gaussian white noise. We derive a theoretical lower bound of the expected improvement achieved in one iteration of CMA-ES, given an estimation of the noise level and the Lipschitz constant of the function's gradient. Solving for the maximum of the lower bound, we obtain a simple expression of the optimal re-evaluation number. We experimentally compare our method to the state-of-the-art noise-handling methods for CMA-ES on a set of artificial test functions across various noise levels, optimization budgets, and dimensionality. Our method demonstrates significant advantages in terms of the probability of hitting near-optimal function values.}, notes = {GECCO-2025 ENUM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(eftimov:2025:GECCO, author = {Tome Eftimov and Peter Korosec}, title = {Adaptive Estimation of the Number of Algorithm Runs in Stochastic Optimization}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Tobias Glasmachers and Youhei Akimoto}, pages = {719--727}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Numerical Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726327}, doi = {doi:10.1145/3712256.3726327}, size = {9 pages}, abstract = {Determining the number of algorithm runs is a critical aspect of experimental design, as it directly influences the experiment's duration and the reliability of its outcomes. This paper introduces an empirical approach to estimating the required number of runs per problem instance for accurate estimation of the performance of the continuous single-objective stochastic optimization algorithm. The method leverages probability theory, incorporating a robustness check to identify significant imbalances in the data distribution relative to the mean, and dynamically adjusts the number of runs during execution as an online approach.The proposed methodology was extensively tested across two algorithm portfolios (104 Differential Evolution configurations and the Nevergrad portfolio) and the COCO benchmark suite, totaling 5,748,000 runs. The results demonstrate 82\%--95\% accuracy in estimations across different algorithms, allowing a reduction of approximately 50\% in the number of runs without compromising optimization outcomes. This online calculation of required runs not only improves benchmarking efficiency, but also contributes to energy reduction, fostering a more environmentally sustainable computing ecosystem.}, notes = {GECCO-2025 ENUM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(girardin:2025:GECCO, author = {Oskar Girardin and Nikolaus Hansen and Dimo Brockhoff and Anne Auger}, title = {Classification-Based Linear Surrogate Modeling of Constraints for {AL-CMA-ES}}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Tobias Glasmachers and Youhei Akimoto}, pages = {728--736}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Numerical Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726435}, doi = {doi:10.1145/3712256.3726435}, size = {9 pages}, abstract = {We introduce linear surrogate functions for modeling inequality constraints to solve constrained blackbox optimization problems with the Augmented Lagrangian CMA-ES. Each surrogate is constructed from a binary classifier that predicts the sign of the constraint value. The classifier, and consequently the resulting algorithm, is invariant under sign preserving transformations of the constraint values and can handle binary, flat, and deceptive constraints. Somewhat surprisingly, we find that adopting a sign-based classification model of the constraints allows to solve classes of constrained problems which can not be solved with the original Augmented Lagrangian method using the true constraint value.}, notes = {GECCO-2025 ENUM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(hamano:2025:GECCO, author = {Ryoki Hamano and Masahiro Nomura and Shota Saito and Kento Uchida and Shinichi Shirakawa}, title = {{CatCMA} with Margin: Stochastic Optimization for Continuous, Integer, and Categorical Variables}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Tobias Glasmachers and Youhei Akimoto}, pages = {737--745}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Numerical Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726471}, doi = {doi:10.1145/3712256.3726471}, size = {9 pages}, abstract = {This study focuses on mixed-variable black-box optimization (MV-BBO), addressing continuous, integer, and categorical variables. Many real-world MV-BBO problems involve dependencies among these different types of variables, requiring efficient methods to optimize them simultaneously. Recently, stochastic optimization methods leveraging the mechanism of the covariance matrix adaptation evolution strategy have shown promising results in mixed-integer or mixed-category optimization. However, such methods cannot handle the three types of variables simultaneously. In this study, we propose CatCMA with Margin (CatCMAwM), a stochastic optimization method for MV-BBO that jointly optimizes continuous, integer, and categorical variables. CatCMAwM is developed by incorporating a novel integer handling into CatCMA, a mixed-category black-box optimization method employing a joint distribution of multivariate Gaussian and categorical distributions. The proposed integer handling is carefully designed by reviewing existing integer handlings and following the design principles of CatCMA. Even when applied to mixed-integer problems, it stabilizes the marginal probability and improves the convergence performance of continuous variables. Numerical experiments show that CatCMAwM effectively handles the three types of variables, outperforming state-of-the-art Bayesian optimization methods and baselines that simply incorporate existing integer handlings into CatCMA.}, notes = {GECCO-2025 ENUM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(nguyen:2025:GECCO, author = {Tuan Anh Nguyen and Ngoc Hoang Luong}, title = {Toward Efficient Mixed-Integer Black-Box Optimization via Evolution Strategies with Plateau Handling Techniques}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Tobias Glasmachers and Youhei Akimoto}, pages = {746--754}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Numerical Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726320}, doi = {doi:10.1145/3712256.3726320}, size = {9 pages}, abstract = {Mixed-Integer Black-Box Optimization (MI-BBO) problems involve optimizing objective functions that have both continuous and integer decision variables without access to any problem-specific knowledge such as gradient or Hessian. Although recent studies have made notable progress in solving many MI-BBO problems, inherent challenges persist, particularly regarding the problem dimensionality and the expected runtime (ERT) required to reach the target values. In this paper, we propose an efficient MI-BBO method that uses evolution strategies with plateau handling techniques. To address problems with higher dimensions, we explore several high-dimensional algorithms, such as VD-CMA (a linear variant of CMA-ES for High Dimension Optimization) and CR-FM-NES (Cost-Reduction Fast Moving Natural Evolution Strategy), and appropriately adapt certain plateau handling techniques to enhance the optimization performance. Numerical experiments with standard benchmark functions against prominent recently-proposed MI-BBO algorithms demonstrate that our methods can solve problems with higher dimensions while maintaining the optimization efficiency. Moreover, the results also reveal the potential for scaling to more challenging problem classes with low computational costs. The source code can be found at: https://github.com/ELO-Lab/eMI-BBO.}, notes = {GECCO-2025 ENUM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(scholman:2025:GECCO, author = {Renzo Scholman and Tanja Alderliesten and Peter Bosman}, title = {More Efficient Real-Valued Gray-Box Optimization through Incremental Distribution Estimation in {RV-GOMEA}}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Tobias Glasmachers and Youhei Akimoto}, pages = {755--763}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {evolutionary algorithms, estimation of distribution algorithms, incremental learning, linkage modeling, Evolutionary Numerical Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726418}, doi = {doi:10.1145/3712256.3726418}, size = {9 pages}, abstract = {The Gene-pool Optimal Mixing EA (GOMEA) family of EAs offers a specific means to exploit problem-specific knowledge through linkage learning, i.e., inter-variable dependency detection, expressed using subsets of variables, that should undergo joint variation. Such knowledge can be exploited if faster fitness evaluations are possible when only a few variables are changed in a solution, enabling large speed-ups. The recent-most version of Real-Valued GOMEA (RV-GOMEA) can learn a conditional linkage model during optimization using fitness-based linkage learning, enabling fine-grained dependency exploitation in learning and sampling a Gaussian distribution. However, while the most efficient Gaussian-based EAs, like NES and CMA-ES, employ incremental learning of the Gaussian distribution rather than performing full re-estimation every generation, the recent-most RV-GOMEA version does not employ such incremental learning. In this paper, we therefore study whether incremental distribution estimation can lead to efficiency enhancements of RV-GOMEA. We consider various benchmark problems with varying degrees of overlapping dependencies. We find that, compared to RV-GOMEA and VKD-CMA-ES, the required number of evaluations to reach high-quality solutions can be reduced by a factor of up to 1.5 if population sizes are tuned problem-specifically, while a reduction by a factor of 2--3 can be achieved with generic population-sizing guidelines.}, notes = {GECCO-2025 ENUM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(sekino:2025:GECCO, author = {Yuta Sekino and Yohei Watanabe and Kento Uchida and Shinichi Shirakawa}, title = {Surrogate-Assisted {CMA-ES} for Problems with Low Effective Dimensionality}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Tobias Glasmachers and Youhei Akimoto}, pages = {764--772}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {CMA-ES, surrogate model, low effective dimensionality, gaussian process regression, Evolutionary Numerical Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726466}, doi = {doi:10.1145/3712256.3726466}, size = {9 pages}, abstract = {High-dimensional optimization problems in real-world applications often possess the property called low effective dimensionality (LED), where only a small part of directions in search space affect the evaluation value, and others are redundant. On problems with LED, because the redundant directions deteriorate the prediction performance of the surrogate model, the performance of several surrogate-assisted evolutionary algorithms is worsened. This paper focuses on the doubly trained surrogate CMA-ES (DTS-CMA-ES) that employs Gaussian process regression as a surrogate model and proposes DTS-CMA-ES-LED by incorporating several countermeasures for LED to DTS-CMA-ES. The proposed method considers directions along the eigenvectors of the covariance matrix and evaluates the effectiveness of each direction using the estimated element-wise signal-to-noise ratio of the update directions. Then, the proposed method reconstructs the kernel function with the computed effectiveness to reduce the effect of redundant directions. We also introduce the hyperparameter adaptation mechanism and refinement of the step-size adaptation as countermeasures for LED. The experimental results show that DTS-CMA-ES-LED effectively optimized the benchmark functions with LED.}, notes = {GECCO-2025 ENUM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(signorelli:2025:GECCO, author = {Federico Signorelli and Anil Yaman}, title = {A Perturbation and Speciation-Based Algorithm for Dynamic Optimization Uninformed of Change}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Tobias Glasmachers and Youhei Akimoto}, pages = {773--781}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Evolutionary Numerical Optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726391}, doi = {doi:10.1145/3712256.3726391}, size = {9 pages}, abstract = {Dynamic optimization problems (DOPs) are challenging due to their changing conditions. This requires algorithms to be highly adaptable and efficient in terms of finding rapidly new optimal solutions under changing conditions. Traditional approaches often depend on explicit change detection, which can be impractical or inefficient when the change detection is unreliable or unfeasible. We propose Perturbation and Speciation-Based Particle Swarm Optimization (PSPSO), a robust algorithm for uninformed dynamic optimization without requiring the information of environmental changes. The PSPSO combines speciation-based niching, deactivation, and a newly proposed random perturbation mechanism to handle DOPs. PSPSO leverages a cyclical multi-population framework, strategic resource allocation, and targeted noisy updates, to adapt to dynamic environments. We compare PSPSO with several state-of-the-art algorithms on the Generalized Moving Peaks Benchmark (GMPB), which covers a variety of scenarios, including simple and multi-modal dynamic optimization, frequent and intense changes, and high-dimensional spaces. Our results show that PSPSO outperforms other state-of-the-art uninformed algorithms in all scenarios and leads to competitive results compared to informed algorithms. In particular, PSPSO shows strength in functions with high dimensionality or high frequency of change in the GMPB. The ablation study showed the importance of the random perturbation component.}, notes = {GECCO-2025 ENUM A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(abdelhafez:2025:GECCO, author = {Amr Abdelhafez and Alexey Lastovetsky}, title = {Energy and Performance Analysis of Parallel Heterogeneous Genetic Algorithms under Various {CPU} and {GPU} {DVFS} Governors: A Preliminary Study on Predictive Profiling}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Dirk Thierens and Elizabeth Wanner}, pages = {782--790}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Genetic Algorithms}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726422}, doi = {doi:10.1145/3712256.3726422}, size = {9 pages}, abstract = {Parallel heterogeneous computing has emerged as a promising approach for addressing computationally intensive problems. Energy efficiency is a critical concern in high-performance computing, particularly when leveraging hybrid architectures such as CPU-GPU systems. This study aims to provide valuable insight into optimizing the trade-off between energy efficiency, performance, and power governors over hybrid architectures.In this work, we evaluate a Parallel Heterogeneous Genetic Algorithm (HPIGA) by running it under five Dynamic Voltage and Frequency Scaling (DVFS) configurations, exploring different frequency configurations for both CPU and GPU. These configurations investigate various combinations of CPU and GPU operating modes, including "powersave" and "performance". Through these experiments, we analyze the energy consumption and performance characteristics of the parallel algorithm under fixed computational loads. The results reveal interesting insights into CPU-GPU specific DVFS configurations, where setting the CPU and GPUs to high/low frequencies can significantly reduce dynamic energy usage in certain configurations.These findings contribute to the development of sustainable computing frameworks by addressing the challenges inherent in frequency scaling and heterogeneous computing environments. This study provides a foundation for future research aimed at developing predictive models and advanced scheduling techniques to further optimize energy efficiency in hybrid CPU/GPU architectures.}, notes = {GECCO-2025 GA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(jiang:2025:GECCO, author = {Yuxin Jiang and Jianyong Sun}, title = {{Evo-SINDy:} Universal Discovery of Partial Differential Equations Using Cooperative Evolutionary Computation}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Dirk Thierens and Elizabeth Wanner}, pages = {791--799}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {data-driven discovery, partial differential equations, multi-population co-evolutionary algorithm, sindy, Genetic Algorithms}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726360}, doi = {doi:10.1145/3712256.3726360}, size = {9 pages}, abstract = {The discovery of the mathematical form of partial differential equations (PDEs) from data has broad applications and significant implications in many fields. Existing data-driven methods such as the well-known SINDy method, however, struggle to identify arbitrary forms of PDEs with minimal prior knowledge. In this paper, we propose a data-driven method for PDE identification, named Evo-SINDy, which leverages a multi-population co-evolutionary algorithm to address the limitations of SINDy. This method is able to efficiently identify PDEs from a sufficiently large search space that best match data characteristics, ensuring minimal reliance on prior knowledge. Experimental results demonstrate that Evo-SINDy can identify more numbers of one-dimensional PDEs within a unified framework than the other known methods, and outperforms two recently-proposed methods that use open libraries in terms of computational efficiency.}, notes = {GECCO-2025 GA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(le:2025:GECCO2, author = {Chi Cuong Le and Tri Phan and Ngoc Hoang Luong}, title = {Gradient-Free Sparse Adversarial Attack on Object Detection Models}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Dirk Thierens and Elizabeth Wanner}, pages = {800--808}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Genetic Algorithms}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726305}, doi = {doi:10.1145/3712256.3726305}, size = {9 pages}, abstract = {The rapid increase of object detection's applications leads to a growing need for these models to be robust to adversarial examples. However, it has been shown that deep neural networks (DNNs) are vulnerable to adversarial examples. In this work, we explore the vulnerability of recent object detection models by generating sparse adversarial examples that differ from the original images by only a few pixels. Moreover, to be suitable for real-world scenarios, we consider the context in which we are ignorant of victim models and employ a gradient-free approach to generate imperceptible adversarial examples. Notably, there are two challenges that we have to address simultaneously: reducing the number of perturbed pixels and limiting the number of queries needed to successfully find an adversarial example. Existing methods usually try to solve only one of those challenges, regardless of the poor quality of the other, and result in high computational resources or perceptible adversarial examples. Our study aims to use only a small number of queries to generate imperceptible perturbations that make object detectors yield wrong predictions. Our experiments are conducted with the convolutional neural network-based YOLO family and the vision transformer-based models (i.e., DINO and DETR) on the PASCAL-VOC dataset.}, notes = {GECCO-2025 GA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(pathirage-don:2025:GECCO, author = {Thilina {Pathirage Don} and Aneta Neumann and Frank Neumann}, title = {Evolutionary Multitasking for the Scenario-based Travelling Thief Problem}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Dirk Thierens and Elizabeth Wanner}, pages = {809--817}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {transfer optimisation, evolutionary multitasking, multifactorial optimisation, travelling thief problem, Genetic Algorithms}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726468}, doi = {doi:10.1145/3712256.3726468}, size = {9 pages}, abstract = {Evolutionary multitasking is an emerging paradigm in optimisation that draws inspiration from cognitive multitasking in humans. It seeks to solve multiple optimisation tasks in parallel referring to a shared population of individuals. This approach uses underlying commonalities between tasks to accelerate the convergence, by leveraging the principles of knowledge transfer. The travelling thief problem (TTP) combines the characteristics of both the travelling salesman problem (TSP) and the 0--1 knapsack problem (KP), depicting the interdependency of multiple components that can be seen in real-world applications. In this study, we represent the TTP problem as a combination of multiple scenarios where each scenario combines a similar TSP component, yet a different KP component. We set up scenarios for the experiments using both fixed weights and weights generated uniformly at random. We follow an evolutionary multitasking optimisation approach to solve multiple scenarios in parallel. By experimenting with a variety of TTP instances, we compare the performance of basic and advanced multitasking approaches, against classical methods. The analysis shows that multitasking brings a competitive advantage over the classical methods when operating on small time budgets.}, notes = {GECCO-2025 GA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(quevedo-de-carvalho:2025:GECCO, author = {Ozeas {Quevedo de Carvalho} and Darrell Whitley}, title = {Dramatically Faster Partition Crossover for the Traveling Salesman Problem}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Dirk Thierens and Elizabeth Wanner}, pages = {818--826}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {traveling salesman problem, combinatorial optimization, genetic algorithms, crossover operators}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726465}, doi = {doi:10.1145/3712256.3726465}, size = {9 pages}, abstract = {The Partition Crossover is a deterministic crossover operator for the Traveling Salesman Problem (TSP). It decomposes the union graph of two TSP solutions, A and B, into connected components known as AB-cycles, from which the lower-cost edges are selected and recombined to produce offspring. The operator finds the best offspring within a search space of 2k solutions in linear time, where k is the number of recombining components. We introduce Generalized Partition Crossover 3 (GPX3), a new implementation of Partition Crossover. GPX3 features a new algorithm to quickly find AB-cycles in the union graph. It also identifies additional recombining AB-cycles, expanding the reachable search space. We show that GPX3 runs in O(n) time and is more efficient and effective than previous implementations of Partition Crossover for the TSP.}, notes = {GECCO-2025 GA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(schmidbauer:2025:GECCO, author = {Marcus Schmidbauer and Dirk Sudholt}, title = {Empirical Linkage Learning Provably Builds Truthful Models on Concatenated Traps and {H-IFF}}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Dirk Thierens and Elizabeth Wanner}, pages = {827--835}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {linkage learning, model-based evolutionary algorithms, parameterless population pyramid, runtime analysis, theory, Genetic Algorithms}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726417}, doi = {doi:10.1145/3712256.3726417}, size = {9 pages}, abstract = {Linkage Learning aims to discover variable dependencies during the optimisation process. To this end, Statistical Linkage Learning (SLL) uses statistical analysis of gene value combinations, whereas Empirical Linkage Learning (ELL) is based on comparing the fitness of neighbouring solutions. ELL, in contrast to SLL, provably does not report false linkage, but is computationally more expensive.We present the first runtime analysis of an ELL-based evolutionary algorithm. Specifically, we analyse the ELL-based Parameter-less Population Pyramid (P3) on concatenated traps and the H-IFF problem, complementing a previous analysis of the SLL-based P3 and thus enabling the theory-founded comparison of linkage-learning techniques. We show that ELL builds models that accurately represent the ground truth based on the currently available information, capturing the maximum possible amount of linkage. This underscores its effectiveness.}, notes = {GECCO-2025 GA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(whitley:2025:GECCO, author = {Darrell Whitley and Gabriela Ochoa and Francisco Chicano}, title = {How Partition Crossover Exposes Parallel Lattices and the Fractal Structure of k-Bounded Functions}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Dirk Thierens and Elizabeth Wanner}, pages = {836--844}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {partition crossover, lattices, genetic algorithms, combinatorial optimization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726340}, doi = {doi:10.1145/3712256.3726340}, size = {9 pages}, abstract = {A combination of recombination and local search can expose the existence of an exponential number of parallel lattices that span the search space for all classes of k-bounded pseudo-Boolean functions, including MAX-kSAT problems. These "parallel" lattices sometimes have identical evaluations shifted by a constant. We use Partition Crossover to aid in the discovery of lattices, which are sets of 2q possible offspring from recombination events, organized into q-dimensional hypercubes, where q is the number of recombining components given two parents. Finally, we show that recursively embedded subspace lattices display a fractal structure, which can be captured using rewrite rules based on a Lindenmayer system that accurately model how local optima are distributed across different size lattices.}, notes = {GECCO-2025 GA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(wigney:2025:GECCO, author = {Liam Wigney and Aneta Neumann and Yew-Soon Ong and Frank Neumann}, title = {On the Use of Matching Algorithms to Transfer Solutions for the Travelling Salesperson Problem}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Dirk Thierens and Elizabeth Wanner}, pages = {845--853}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {evolutionary multitasking, transfer learning, travelling salesperson problem, theory, Genetic Algorithms}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726469}, doi = {doi:10.1145/3712256.3726469}, size = {9 pages}, abstract = {Multitasking evolutionary algorithms can be effectively used to solve a number of problems with a single population. A key issue in deciding their effectiveness, is how to transfer good solutions from one problem instance to another problem instance which shares some characteristics. We investigate in this paper how to transfer solutions between different problem instances of the Travelling Salesperson Problem (TSP) based matching algorithms and introduce different transfer mechanisms based on matching the nodes between problem instances. In our experimental study, we examine how the different transfer approaches perform for different classes of TSP instances dependent on the characteristics of the considered problem instances.}, notes = {GECCO-2025 GA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(zychowski:2025:GECCO, author = {Adam Zychowski and Abhishek Gupta and Yew Soon Ong and Jacek Mandziuk}, title = {Augmented Decision Spaces for Stackelberg Security Games: Sparse evolution begets scalability}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Alberto Moraglio and James McDermott}, pages = {854--862}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {General Evolutionary Computation, Hybrids}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726440}, doi = {doi:10.1145/3712256.3726440}, size = {9 pages}, abstract = {This paper introduces the Augmented Decision Space Optimization (ADSO) method for sparsity-driven optimization of mixed-strategies in Stackelberg Security Games (SSGs). The proposed method enhances traditional strategy optimization by combining binary variables to represent the presence of pure strategies with real-valued variables to refine their selection probabilities. Specifically, instead of waiting for an evolutionary process to gradually discover sparse solutions, the binary variables in ADS allow the real-valued variables to be switched on or off, thereby directly enforcing sparsity. This dual codification scheme achieves targets such as sparsification and computational efficiency in large-scale games. We demonstrate that ADS outperforms existing heuristic methods, offering superior solution quality, scalability, and stability. Empirical results across three different benchmark games show that ADS generates compact strategies with minimal computational overhead, achieving performance close to the exact methods. Furthermore, state-of-the-art results are obtained for problems where exact methods fail to scale effectively. Our framework promises broad applicability beyond SSGs, encompassing a wide range of game-theoretic and combinatorial optimization problems.}, notes = {GECCO-2025 GECH A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(zychowski:2025:GECCO2, author = {Adam Zychowski and Xin Yao and Jacek Mandziuk}, title = {Diversity-driven Cooperating Portfolio of Metaheuristic Algorithms}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Alberto Moraglio and James McDermott}, pages = {863--871}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {General Evolutionary Computation, Hybrids}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726442}, doi = {doi:10.1145/3712256.3726442}, size = {9 pages}, abstract = {The paper introduces a novel hybrid island-based framework in which diverse metaheuristics cooperate to effectively explore the search space. A core component of the framework is a diversity-driven migration mechanism, enabling adaptive management of the information flow between islands. Three fundamental aspects of migration - what to migrate, when to migrate, and where to migrate - are thoroughly analyzed, leading to the development of strategies that foster synergy between heterogeneous algorithms. These strategies balance exploration and exploitation, ensuring effective global and local search. The framework was evaluated on a set of diverse optimization benchmarks, both discrete (Traveling Salesman Problem instances) and continuous (BBOB functions). Experimental results demonstrate that the proposed approach surpasses traditional algorithms and their island-based variants in convergence speed, solution quality, and resilience to stagnation. Adaptive mechanisms dynamically adjust migration strategies during the optimization process, further enhancing the framework's effectiveness. The proposed method represents an advancement in hybrid metaheuristic systems, offering scalability and flexibility that are essential for solving complex optimization tasks.}, notes = {GECCO-2025 GECH A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(baumann:2025:GECCO, author = {Jakob Baumann and Ignaz Rutter and Dirk Sudholt}, title = {Analysing the Effectiveness of Mutation Operators for One-Sided Bipartite Crossing Minimisation}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Alberto Moraglio and James McDermott}, pages = {872--880}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {graph drawing, network layout, permutation spaces, mutation operators, runtime analysis, theory, hybridisation, General Evolutionary Computation, Hybrids}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726407}, doi = {doi:10.1145/3712256.3726407}, size = {9 pages}, abstract = {Graph Drawing aims to make graphs visually comprehensible while faithfully representing their structure. In layered drawings, each vertex is drawn on a horizontal line and edges are drawn as y-monotone curves. We consider a fundamental problem from this domain, the One-Sided Bipartite Crossing Minimisation (OBCM) problem. Given a bipartite graph with two layers and a fixed horizontal order of vertices on the first layer, the objective is to order the vertices on the second layer to minimise the number of edge crossings.We empirically analyse the performance of simple evolutionary algorithms (EAs) for OBCM and compare different mutation operators for the underlying permutation problem: exchanging two elements (exchange), swapping adjacent elements (swap) and jumping an element to a new position (jump). Our analysis reveals that jump is the most effective operator, with EAs using jumps outperforming all classical algorithms in terms of solution quality within a reasonable number of generations. We also propose hybrid EA variants that reduce the required number of generations by up to a factor of 100. Additionally, we provide theoretical insights and prove a quadratic upper bound on the expected runtime for the most effective EA using jumps for a general class of instances.}, notes = {GECCO-2025 GECH A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(corus:2025:GECCO, author = {Dogan Corus and Pietro S. Oliveto and Feiyang Zheng}, title = {Hybrid Selection Allows Steady-State Evolutionary Algorithms to Control the Selective Pressure in Multimodal Optimisation}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Alberto Moraglio and James McDermott}, pages = {881--889}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {populations, selective pressure, hybridisation, self-adaptation, General Evolutionary Computation, Hybrids}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726411}, doi = {doi:10.1145/3712256.3726411}, size = {9 pages}, abstract = {Recent work has shown that Inverse Tournament Selection operators within steady-state evolutionary algorithms (EAs) allow to control the selective pressure much more accurately than in generational EAs. However, to achieve low selective pressures, large tournament sizes are required which come at the cost of prohibitive expected times for the population to escape from local optima. To this end, we propose a hybrid selection mechanism that leads to considerable speed-ups in the expected time to escape from local optima while permitting to keep the selective pressure arbitrarily low and the use of large population sizes. The mechanism simply switches between Inverse Elitist selection and Uniform selection when it detects that the population is stuck on local optima, and switches back when an improving solution is found. We prove its effectiveness for the TruncatedTwomaxk and RidgeWithBranchesj benchmarks from the literature by providing super-linear speed-ups over the (mu+1) EA with any fixed selective pressure.}, notes = {GECCO-2025 GECH A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(de-queiroz:2025:GECCO, author = {Thiago Alves {de Queiroz} and Manuel Iori and Alberto Locatelli and Matthieu Parizy}, title = {Solving the Cubic Knapsack Problem using the Quantum-Inspired Digital Annealer Technology}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Alberto Moraglio and James McDermott}, pages = {890--897}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {cubic knapsack problem, quantum computing, digital annealer, HUBO, QUBO, General Evolutionary Computation, Hybrids}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726474}, doi = {doi:10.1145/3712256.3726474}, size = {8 pages}, abstract = {This study investigates the effectiveness of quantum methods in tackling the cubic knapsack problem (CKP). The CKP is not only NP-hard but also extremely difficult to solve in practice. Benchmark instances of small size (including some with only 60 items) remain unsolved to proven optimality. We solve the CKP using the latest Digital Annealer (DA) prototype, an extended Ising machine available through the Quantum-Inspired Integrated Optimization (QIIO) service on Fujitsu's Kozuchi platform. Specifically, we propose two formulations: a higher-order unconstrained binary optimization (HUBO) and a quadratic unconstrained binary optimization. The latter is derived by reformulating the HUBO model into an equivalent quadratic form. These models are solved using the QIIO solver and compared with three state-of-the-art algorithms, a greedy heuristic, and two mixed integer programs. Additionally, we introduce a postprocessing heuristic to ensure the feasibility of solutions generated by the DA solver, as within short time limits, it does not always produce feasible solutions. Computational experiments are conducted on instances with up to 200 items and varying densities of nonzero objective coefficients. The results indicate that the HUBO formulation is highly competitive with state-of-the-art algorithms, achieving the best new solutions for six large instances.}, notes = {GECCO-2025 GECH A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(komosinski:2025:GECCO, author = {Maciej Komosinski and Agnieszka Mensfelt}, title = {Enhancing Quality-Diversity Optimization Through Domain-Specific Dissimilarity as Crowding Distance}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Alberto Moraglio and James McDermott}, pages = {898--906}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {General Evolutionary Computation, Hybrids}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726459}, doi = {doi:10.1145/3712256.3726459}, size = {9 pages}, abstract = {Quality-diversity algorithms aim to simultaneously optimize solution performance and maintain diversity within a population. In this paper, we explore the use of NSGA-II as a quality-diversity algorithm for the evolutionary design of 3D structures, modifying its crowding distance calculation to use dissimilarity measures. While NSGA-II is widely employed for multi-objective optimization, its use of fitness for calculating crowding distance may not be the most effective for tasks requiring solution diversity. We propose leveraging both genetic and phenotypic dissimilarity metrics to improve diversity management. To evaluate this approach, we compare the standard NSGA-II using fitness-based crowding distance and Diversity-Enhancing NSGA-II (DE-NSGA-II) using various combinations of dissimilarity-based metrics for crowding distance and diversity scores. Experiments are conducted using two distinct genetic representations on two optimization tasks: height of the center of gravity of passive structures and velocity of active structures. Results demonstrate the potential of dissimilarity-based crowding distance to enhance the diversity and overall quality of solutions in complex evolutionary design tasks.}, notes = {GECCO-2025 GECH A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(koelle:2025:GECCO, author = {Michael Koelle and Tom Bintener and Maximilian Zorn and Gerhard Stenzel and Leo Suenkel and Thomas Gabor and Claudia Linnhoff-Popien}, title = {Evaluating Mutation Techniques in Genetic-Algorithm-Based Quantum Circuit Synthesis}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Alberto Moraglio and James McDermott}, pages = {907--915}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {variational quantum circuits, automated circuit design, mutation, General Evolutionary Computation, Hybrids}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726402}, doi = {doi:10.1145/3712256.3726402}, size = {9 pages}, abstract = {Quantum computing leverages the unique properties of qubits and quantum parallelism to solve problems intractable for classical systems, offering unparalleled computational potential. However, optimization of quantum circuits remains critical, especially for noisy intermediate-scale quantum (NISQ) devices with limited qubits and high error rates. Genetic algorithms (GAs) provide a promising approach for efficient quantum circuit synthesis by automating optimization tasks. This work examines the impact of various mutation strategies within a GA framework for quantum circuit synthesis. By analyzing how different mutations transform circuits, it identifies strategies that enhance efficiency and performance. Experiments used a fitness function emphasizing fidelity, while accounting for circuit depth and T-operations, to optimize circuits with four to six qubits. Our analysis revealed that, while the "swap, addition" strategy achieved the highest fidelity scores, it consistently increased circuit depth. In contrast, combining "swap, addition, delete" strategies offers a more balanced approach, delivering near-optimal results while also having the potential of reducing circuit depth.}, notes = {GECCO-2025 GECH A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(leite:2025:GECCO, author = {Rui Leite and Hernan Aguirre and Kiyoshi Tanaka}, title = {Key Insights into Estimating Nash Equilibria in Simultaneous Continuous Multiplayer Games Using Coevolutionary Algorithms}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Alberto Moraglio and James McDermott}, pages = {916--924}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {multi-objective optimization, game theory, coevolution, mutation operators, security, cybersecurity and defense, General Evolutionary Computation, Hybrids}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726373}, doi = {doi:10.1145/3712256.3726373}, size = {9 pages}, abstract = {Game theory is a powerful tool for analyzing strategic interactions between rational agents and has been widely applied across fields such as economics, biology, and cybersecurity. In this paper, we propose a novel approach for estimating solutions to multiplayer games of simultaneous decision with continuous strategy sets, including those with infinitely many Nash Equilibria. Our method leverages the coevolution of multiple Evolutionary Algorithms (EAs): a single-objective EA models a single-objective player, while a Pareto dominance-based EA represents a multi-objective player. Each EA optimizes its player's strategies (decisions) through iterative gameplay. We analyze the key features that enable the proposed algorithm to estimate a Nash Equilibrium with minimal deviation from the analytical solution (which remains unknown to the algorithm) and to maintain stability near this solution. Experimental results show that the proposed algorithm converges to the nearest equilibrium with appropriate parameter tuning, including the secondary parent/survival selection criterion for the multi-objective EA, the fitness computation method, the mutation distribution index, and the mutation rate.}, notes = {GECCO-2025 GECH A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(lerasle:2025:GECCO, author = {Matthieu Lerasle and Abderrahim Bendahi and Adrien Fradin}, title = {Unlearning Works Better Than You Think: Local Reinforcement-Based Selection of Auxiliary Objectives}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Alberto Moraglio and James McDermott}, pages = {925--933}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {General Evolutionary Computation, Hybrids}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726380}, doi = {doi:10.1145/3712256.3726380}, size = {9 pages}, abstract = {We introduce Local Reinforcement-Based Selection of Auxiliary Objectives (LRSAO), a novel approach that selects auxiliary objectives using reinforcement learning (RL) to support the optimization process of an evolutionary algorithm (EA) as in EA+RL framework and furthermore incorporates the ability to unlearn previously used objectives. By modifying the reward mechanism to penalize moves that do no increase the fitness value and relying on the local auxiliary objectives, LRSAO dynamically adapts its selection strategy to optimize performance according to the landscape and unlearn previous objectives when necessary.We analyze and evaluate LRSAO on the black-box complexity version of the non-monotonic Jumpl function, with gap parameter l, where each auxiliary objective is beneficial at specific stages of optimization. The Jumpl function is hard to optimize for evolutionary-based algorithms and the best-known complexity for reinforcement-based selection on Jumpl was O(n2 log(n)/l). Our approach improves over this result to achieve a complexity of Theta(n2/l2 + n log(n)) resulting in a significant improvement, which demonstrates the efficiency and adaptability of LRSAO, highlighting its potential to outperform traditional methods in complex optimization scenarios.Code is available at https://github.com/FAdrien/LRSAO.}, notes = {GECCO-2025 GECH A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(suenkel:2025:GECCO, author = {Leo Suenkel and Philipp Altmann and Michael Koelle and Gerhard Stenzel and Thomas Gabor and Claudia Linnhoff-Popien}, title = {Quantum Circuit Construction and Optimization through Hybrid Evolutionary Algorithms}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Alberto Moraglio and James McDermott}, pages = {934--942}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {General Evolutionary Computation, Hybrids}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726392}, doi = {doi:10.1145/3712256.3726392}, size = {9 pages}, abstract = {We apply a hybrid evolutionary algorithm to minimize the depth of circuits in quantum computing. More specifically, we evaluate two different variants of the algorithm. In the first approach, we combine the evolutionary algorithm with an optimization subroutine to optimize the parameters of the rotation gates present in the quantum circuit. In the second, the algorithm solely relies on evolutionary operations (i.e., mutations and crossover). We approach the problem from two sides: (1) constructing circuits from the ground up by starting with random initializations and (2) initializing individuals with a target circuit in order to optimize it further according to the fitness function. We run experiments on random circuits with 4 and 6 qubits varying in circuit depth. Our results show that the proposed methods are able to significantly reduce the depth of circuits while still retaining a high fidelity to the target state.}, notes = {GECCO-2025 GECH A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(van-stein:2025:GECCO, author = {Niki {van Stein} and Anna {V. Kononova} and Lars Kotthoff and Thomas Baeck}, title = {Code Evolution Graphs: Understanding Large Language Model Driven Design of Algorithms}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Alberto Moraglio and James McDermott}, pages = {943--951}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {General Evolutionary Computation, Hybrids}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726328}, doi = {doi:10.1145/3712256.3726328}, size = {9 pages}, abstract = {Large Language Models (LLMs) have demonstrated great promise in generating code, especially when used inside an evolutionary computation framework to iteratively optimize the generated algorithms. However, in some cases they fail to generate competitive algorithms or the code optimization stalls, and we are left with no recourse because of a lack of understanding of the generation process and generated codes. We present a novel approach to mitigate this problem by enabling users to analyze the generated codes inside the evolutionary process and how they evolve over repeated prompting of the LLM. We show results for three benchmark problem classes and demonstrate novel insights. In particular, LLMs tend to generate more complex code with repeated prompting, but additional complexity can hurt algorithmic performance in some cases. Different LLMs have different coding "styles" and generated code tends to be dissimilar to other LLMs. These two findings suggest that using different LLMs inside the code evolution frameworks might produce higher performing code than using only one LLM.}, notes = {GECCO-2025 GECH A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(anthes:2025:GECCO, author = {Philipp Anthes and Dominik Sobania and Franz Rothlauf}, title = {Transformer Semantic Genetic Programming for Symbolic Regression}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {952--960}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, transformer models, semantic operators, symbolic regression}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726412}, doi = {doi:10.1145/3712256.3726412}, size = {9 pages}, abstract = {In standard genetic programming (stdGP), solutions are varied by modifying their syntax, with uncertain effects on their semantics. Geometric-semantic genetic programming (GSGP), a popular variant of GP, effectively searches the semantic solution space using variation operations based on linear combinations, although it results in significantly larger solutions. This paper presents Transformer Semantic Genetic Programming (TSGP), a novel and flexible semantic approach that uses a generative transformer model as search operator. The transformer is trained on synthetic test problems and learns semantic similarities between solutions. Once the model is trained, it can be used to create offspring solutions with high semantic similarity also for unseen and unknown problems. Experiments on several symbolic regression problems show that TSGP generates solutions with comparable or even significantly better prediction quality than stdGP, SLIM_GSGP, DSR, and DAE-GP. Like SLIM_GSGP, TSGP is able to create new solutions that are semantically similar without creating solutions of large size. An analysis of the search dynamic reveals that the solutions generated by TSGP are semantically more similar than the solutions generated by the benchmark approaches allowing a better exploration of the semantic solution space.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(bai:2025:GECCO, author = {Minghui Bai and Xiaoying Gao and Jiaxin Niu and Jianbin Ma}, title = {Multi-Objective Genetic Programming for Imbalanced Classification with Adaptive Thresholds and a New Fitness Function}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {961--969}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, multi-objective, threshold, classifier, imbalanced}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726348}, doi = {doi:10.1145/3712256.3726348}, size = {9 pages}, abstract = {Genetic programming (GP) is widely used for classifier construction due to its flexible representation and feature construction characteristics. Traditional GP methods, however, often rely on a fixed threshold, typically 0, which fails to reflect the true distribution of the data in imbalanced datasets. To overcome this, we propose a multi-objective GP method that adaptively adjusts the threshold during evolution using Youden's Index. This adaptive threshold adjustment allows the classifiers to better fit the data distribution. Additionally, we introduce a class separation metric, distt, aimed at enhancing the clarity of the classification boundaries and improving the generalization ability of the evolved classifiers. We use the multi-objective GP, along with the optimal threshold of each classifier, to jointly optimize the accuracy of the minority and majority classes, as well as the class separation metric distt, selecting the best classifier from the Pareto front for unseen data. Experiments on 7 imbalanced datasets demonstrate that our method outperforms single-objective GP with fixed thresholds and four GP-based algorithms, showcasing superior performance and improved classification clarity. Furthermore, our proposed clarity metric distt improves classification performance, ensuring better generalization and enhanced decision boundaries.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(bakurov:2025:GECCO, author = {Illya Bakurov and Aidan Murphy and Charles Ofria and Wolfgang Banzhaf}, title = {A comparison of tournament and lexicase selection paradigms in regression problems: error-based fitness versus correlation fitness}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {970--979}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726448}, doi = {doi:10.1145/3712256.3726448}, size = {10 pages}, abstract = {Lexicase parent selection considers training cases separately, postulating that aggregated fitness reduces the information about the behavior of individuals. Originally lexicase was proposed in the context of program synthesis, characterized by uncompromising problems that require qualitatively different actions for different inputs, but it has since been extended to regression problems. To facilitate valley-crossing a relaxation parameter epsilon was added broadening the pass condition at a given training case. Although epsilon-lexicase has demonstrated superior effectiveness, it was compared against selection methods that aggregated squared (or absolute) errors. Recent contributions, however, demonstrate that correlation fitness functions can lead to significant performance gains over the root mean square error (RMSE) in tournament-guided evolution for symbolic regression. Here we compare epsilon-lexicase (with and without down-sampling) against tournament selection using both error- and correlation-based fitness to guide Genetic Programming (GP). We also assess batch epsilon-lexicase selection as an intermediate condition. Finally, we explore different selection pressures to assess the exploration-exploitation trade-off. We analyze the experimental results using different metrics, including code redundancy, sharpness-awareness and selection impact. Our results demonstrate that tournament selection with correlation fitness function significantly outperforms epsilon-lexicase on regression problems and that its batch variant also benefits from correlation-based aggregation.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(craine:2025:GECCO, author = {Benjamin Craine and Barry Porter}, title = {Uniform Projection of Program Space Geometry for Genetic Improvement of Software}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {980--988}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, genetic improvement}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726393}, doi = {doi:10.1145/3712256.3726393}, size = {9 pages}, abstract = {Current Genetic Improvement (GI) for software systems use preexisting program representations, such as abstract syntax trees and bytecode, to apply genetic operations to. These representations, however, were designed for the purpose of translating human readable source code to machine code. When used to underpin GI, these representations have drawbacks, such as the risk of breaking a program when deploying mutations. We present a novel matrix-based program representation which is specifically designed for the purpose of GI. Our representation (i) makes it impossible for mutations or crossover to yield an invalid program, without the need for any syntactic or semantic checks, while still making every valid program reachable by search, and (ii) supports the simple expression of rich, layered probability distributions atop the program matrix to guide a GI search process. We build an end-to-end GI system using this new representation and demonstrate how we can layer a range a probability distribution on top of the representation to gain different effects. We also explore the future research possibilities that this approach to program representation presents.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(de-franca:2025:GECCO2, author = {Fabricio Olivetti {de Franca} and Gabriel Kronberger}, title = {Improving Genetic Programming for Symbolic Regression with Equality Graphs}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {989--998}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726383}, doi = {doi:10.1145/3712256.3726383}, size = {10 pages}, abstract = {The search for symbolic regression models with genetic programming (GP) has a tendency of revisiting expressions in their original or equivalent forms. Repeatedly evaluating equivalent expressions is inefficient, as it does not immediately lead to better solutions. However, evolutionary algorithms require diversity and should allow the accumulation of inactive building blocks that can play an important role at a later point. The equality graph is a data structure capable of compactly storing expressions and their equivalent forms allowing an efficient verification of whether an expression has been visited in any of their stored equivalent forms. We exploit the e-graph to adapt the subtree operators to reduce the chances of revisiting expressions. Our adaptation, called eggp, stores every visited expression in the e-graph, allowing us to filter out from the available selection of subtrees all the combinations that would create already visited expressions. Results show that, for small expressions, this approach improves the performance of a simple GP algorithm to compete with PySR and Operon without increasing computational cost. As a highlight, eggp was capable of reliably delivering short and at the same time accurate models for a selected set of benchmarks from SRBench and a set of real-world datasets.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(doz:2025:GECCO, author = {Romina Doz and Francesca Randone and Eric Medvet and Luca Bortolussi}, title = {Evolutionary Synthesis of Probabilistic Programs}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {999--1007}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, grammatical evolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726388}, doi = {doi:10.1145/3712256.3726388}, size = {9 pages}, abstract = {Modeling the relationships between variables through probability distributions lies at the core of probabilistic models, enabling reasoning under uncertainty. Probabilistic programming offers an effective way to represent these models by blending the simplicity of standard programming constructs with the power of automatic inference algorithms. The languages for expressing probabilistic programs are augmented with primitives representing various probability distributions to effectively capture the stochastic behavior inherent in the data. However, writing a probabilistic program is hard, because it typically requires prior knowledge about the data generation mechanism. In this work, we propose a framework for automatically synthesizing probabilistic programs directly from data, thereby learning the underlying relationships between variables and the data-generating process. We adopt an evolutionary approach, specifically grammatical evolution (GE), to extensively explore the space of probabilistic programs, aiming to discover the most likely program that describes the observed data. We experimentally evaluate our method across several benchmarks, incorporating varying levels of prior knowledge through a sketching strategy embedded into the grammar fed to GE, to demonstrate the potential of this evolutionary framework. This evaluation highlights the flexibility and effectiveness of GE in synthesizing probabilistic programs under different informational constraints.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(hu:2025:GECCO, author = {Ting Hu and Wolfgang Banzhaf and Gabriela Ochoa}, title = {How Neutrality Shapes Evolution: Simplicity Bias and Search}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {1008--1016}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726454}, doi = {doi:10.1145/3712256.3726454}, size = {9 pages}, abstract = {Neutrality, characterized by pathways in the genotype space that do not alter the phenotype or fitness, enables a broad exploration of evolutionary search. Simplicity bias describes the tendency of evolutionary systems to favor low-complexity solutions. This study investigates how neutrality contributes to simplicity bias in evolutionary systems using a Boolean Linear Genetic Programming framework. We introduce two fitness functions that use symmetry in solutions to promote neutrality, to analyze their effects on neutral network connectivity and search dynamics. Our results demonstrate that simpler phenotypes, characterized by lower Kolmogorov complexity, exhibit greater redundancy and connectivity, making them more accessible during neutral exploration. In addition, the proposed fitness functions significantly improve search success rates, especially for complex target phenotypes, by expanding neutral pathways. These findings shed light on the role of neutrality in shaping simplicity bias and provide practical insights to improve the effectiveness of evolutionary algorithms.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(nemeth:2025:GECCO, author = {Zsolt Nemeth and Penn {Faulkner Rainford} and Barry Porter}, title = {Reaching Meaningful Diversity with Speciation-Novelty in Genetic Improvement for Software}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {1017--1025}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, genetic improvement, genetic diversity, speciation, novelty search, fitness landscape}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726366}, doi = {doi:10.1145/3712256.3726366}, size = {9 pages}, abstract = {Genetic Improvement (GI) for software has been used in automated bug fixing and in automated performance improvement. Automated improvement has been targeted at multi-context problems, where one implementation variant might be best at one context, and another might be best at a different context. However, this application of GI generally requires a fresh improvement process for each new context, which can be computationally expensive. We propose a novel application of GI for multi-context problems, in which we aim for a diverse set of individuals in an initial training run for one context. We use a phenotypic speciation metric as a diversity indicator, allowing us to plot a diversity geometry through program search space. When a different context is introduced, as a new optimisation target for GI, we are able to select from one of these diverse individuals as a close starting point for fine-tuning. With a hash table implementation as an example to genetically improve, we show that we can exercise a high degree of control over population diversity, and that this diversity can be a useful starting point for finding individuals in successive alternative contexts.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(rosenfeld:2025:GECCO, author = {Liah Rosenfeld and Davide Farinati and Diogo Rasteiro and Gloria Pietropolli and Karina {Brotto Rebuli} and Sara Silva and Leonardo Vanneschi}, title = {Slim-gsgp: A Python Library for Non-Bloating {GSGP}}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {1026--1034}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726398}, doi = {doi:10.1145/3712256.3726398}, code_url = {https://github.com/DALabNOVA/slim}, size = {9 pages}, abstract = {This paper presents slim_gsgp: an open-source Python library that provides the first ever framework for the Semantic Learning algorithm based on Inflate and deflate Mutation (SLIM-GSGP). Proposed in 2024, SLIM-GSGP is a promising non-bloating variant of Geometric Semantic Genetic Programming (GSGP). slim_gsgp includes all existing SLIM-GSGP variants, as well as traditional GSGP and standard Genetic Programming (GP), facilitating comparative analysis and benchmarking. Additionally, slim_gsgp's parallel computation and semi-modular architecture renders it not only fast but also user-friendly and easily extensible, thereby serving as a valuable resource for researchers aiming to advance this emerging and promising area of research. The source code and documentation can be accessed at https://github.com/DALabNOVA/slim.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(shi:2025:GECCO, author = {Jiaming Shi and Kei Sen Fong and Mehul Motani}, title = {Analysis of Memory-Runtime Trade-offs in Caching Strategies for Genetic Programming Symbolic Regression}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {1035--1043}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, symbolic regression, cache strategies}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726375}, doi = {doi:10.1145/3712256.3726375}, size = {9 pages}, abstract = {Genetic Programming Symbolic Regression (GPSR) generates mathematical expressions to model input-output relationships using an evolutionary process. A significant challenge in GPSR lies in the repeated evaluation of entire expressions or their sub-expression, which inflates computational runtime. To address this inefficiency, caching mechanisms have been employed to reduce redundant computations. However, prior studies predominantly employ a single caching strategy, offering limited insights into their comparative performance or memory-runtime trade-offs. In this paper, we present a comprehensive analysis of caching mechanisms for GPSR on synthetic and real-world datasets. We also include an empirical study of key-value usage frequencies under an infinitely large cache, offering insights into optimal cache sizing. Furthermore, we provide actionable guidelines for configuring caching strategies based on computational and memory constraints. Our findings indicate that complex caching mechanisms necessitate a minimum cache size to achieve computational time reductions. Conversely, lightweight caching strategies, such as Least Recently Used (LRU) and, notably, First-In-First-Out (FIFO), can significantly decrease computation time for fitness evaluations, which are a substantial component of the overall runtime.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(simoes:2025:GECCO, author = {Jose Maria Simoes and Nuno Lourenco and Penousal Machado}, title = {Desire-Driven Selection: An Epigenetic Experiment in Genetic Programming}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {1044--1052}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, parent selection, sexual selection, mate choice}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726358}, doi = {doi:10.1145/3712256.3726358}, size = {9 pages}, abstract = {In nature, survival poses small benefits if one fails to reproduce and spread one's genes. This is particularly relevant in sexually reproductive species, which exerts another pressure dimension on the individual beyond natural selection: Sexual Selection. More often than not, the quality of the chosen mate is a crucial step in reproduction, making all the investment in mate choice worthwhile. This partly explains why partners often prefer certain secondary traits, such as ornaments, particularly if such traits signal good fitness. We hypothesize that the dynamics between mating preferences and fitness-dependent ornaments can act as a filter to find a mate within a population, exploiting good solutions while maintaining high diversity. In this work, we propose a new selection method for Genetic Programming based on these premises, validating our approach on regression problems. Results show that high levels of diversity are maintained when compared against a standard tournament selection with performance gains, reducing the overall error by 16.3\% and 13.8\% in training and testing respectively, and performing up to par with state-of-the-art Lexicase selection while also providing the best overall solution.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(tran:2025:GECCO, author = {Binh Tran and Su Nguyen}, title = {An Online Genetic Programming Approach to Dynamic Production Scheduling}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {1053--1061}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, hyper-heuristic, online machine learning}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726342}, doi = {doi:10.1145/3712256.3726342}, size = {9 pages}, abstract = {Due to the production system's complexity and dynamic changes, designing such scheduling strategies is challenging. Recently, advanced machine learning and optimisation methods such as genetic programming (GP) have shown promise in designing sophisticated scheduling strategies. These methods' success relies on accurate data-driven simulation models for evaluating automatically-generated scheduling strategies. However, building a simulation model that accurately predicts complex production system behaviours requires a lot of historical operational data, which may not always be available, especially for new production systems or those adaptive to the market. To overcome this limitation, this study develops the first online GP method called OGP for dynamic production scheduling problems that allows GP to learn and optimise scheduling decisions on the fly without an exact model for fitness evaluations. The experiments with dynamic flexible job shops show that OGP outperforms existing scheduling strategies in the literature when both scheduling and routing decisions are considered. When used as an automated heuristic design method, OGP can generate competitive rules compared to the state-of-the-art GP methods in terms of test performance and rule sizes.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(vacher:2025:GECCO, author = {Quentin Vacher and Stephen Kelly and Ali Naqvi and Nicolas Beuve and Tanya Djavaherpour and Mickael Dardaillon and Karol Desnos}, title = {{MAPLE:} Multi-Action Programs through Linear Evolution for Continuous Multi-Action Reinforcement Learning}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {1062--1071}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, reinforcement learning, interpretability}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726400}, doi = {doi:10.1145/3712256.3726400}, size = {10 pages}, abstract = {Over the last decades, the need to solve complex tasks using machine learning techniques has grown significantly. Deep learning algorithms achieve state-of-the-art performance in most tasks, but at the cost of high computational complexity and limited interpretability. In domains such as Reinforcement Learning (RL), understanding the agent behavior ensures reliability and safety. In this work, we explore Genetic Programming (GP) as a promising solution for RL tasks, providing simpler and more interpretable solutions. While GP achieves competitive results in low-complexity environments, it struggles in environments with high-dimensional action spaces. To address this, we propose Multi-Action Programs through Linear Evolution (MAPLE), a GP algorithm in which the agent is a team of multiple Linear Genetic Programs (LGPs), each responsible for an action. MAPLE is evaluated on the MuJoCo suite and outperforms state-of-the-art GP algorithms and a small deep RL model. It achieves comparable performance to a larger deep RL network in low-dimensional environments while maintaining significantly lower complexity. By decomposing the action decision into different programs, it is possible to understand which parts of the states are needed for each action. This demonstrates the potential of MAPLE for interpretable and efficient solutions in RL.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(vella-zarb:2025:GECCO, author = {David {Vella Zarb} and Geoff Parks and Timoleon Kipouros}, title = {Program Synthesis with {LLM-Predicted} Minimal Specialized Grammars}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {1072--1080}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, grammatical evolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726430}, doi = {doi:10.1145/3712256.3726430}, size = {9 pages}, abstract = {Methods using recent advances in large language models and genetic programming have outperformed GE on some benchmarks, affirming the need for scalable approaches. This paper addresses the scalability challenge by proposing an in-context learning method to automatically generate minimal specialized grammars (MSGs), which are problem-specific subsets of a larger grammar designed to reduce search space complexity. To the best of our knowledge, this represents the first use of in-context learning for program synthesis within GE. Our approach conditions a language model on examples of Backus-Naur Form grammars to generate MSGs tailored to individual synthesis tasks. We evaluate this framework on a benchmark suite widely used in GE research. Experimental results show our method almost always outperforms the baseline GE approach, improving both the number and frequency of problems solved while reducing computational cost as measured by fitness evaluations.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(vitel:2025:GECCO, author = {Dmytro Vitel and Kok Cheng Tan and Alessio Gaspar and Paul Wiegand}, title = {Coordinate System Extraction as the Search Driver in Test-Based Genetic Programming}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {1081--1089}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, test-based genetic programming, interaction matrix coordinate system, underlying objectives, derived objectives, many-objective optimization, selection for breeding, coevolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726464}, doi = {doi:10.1145/3712256.3726464}, size = {9 pages}, abstract = {In test-based genetic programming (GP), the evolution is driven by program-test interactions that are naturally multidimensional. Previous works applied dimension reductions to these interaction matrices to form "derived objectives" that guide evolutionary multiobjective optimization (EMO). In this work, we consider tests as separate optimization targets as an alternative to reducing the interaction dimensionality. We compare methods based on different Pareto-front sampling strategies and propose a coevolutionary approach driven by a selection method based on extracting the underlying game structure from the interactions. This structure is a multidimensional coordinate system that maintains domination relations between programs along the axes and facilitates better sampling for breeding. Experimental results in discrete value domains demonstrate that the proposed methods have, in many cases, better performance on benchmarks than methods based on fitness aggregation, including dimensionality reduction.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(xu:2025:GECCO2, author = {Meng Xu and Frank Neumann and Aneta Neumann and Yew Soon Ong}, title = {Quality Diversity Genetic Programming for Learning Scheduling Heuristics}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Aniko Ekart and Nelishia Pillay}, pages = {1090--1098}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, quality and diversity optimization, dynamic flexible job shop scheduling}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726343}, doi = {doi:10.1145/3712256.3726343}, size = {9 pages}, abstract = {Real-world optimization often demands diverse, high-quality solutions. Quality-Diversity (QD) optimization is a multifaceted approach in evolutionary algorithms that aims to generate a set of solutions that are both high-performing and diverse. QD algorithms have been successfully applied across various domains, providing robust solutions by exploring diverse behavioral niches. However, their application has primarily focused on static problems, with limited exploration in the context of dynamic combinatorial optimization problems. Furthermore, the theoretical understanding of QD algorithms remains underdeveloped, particularly when applied to learning heuristics instead of directly learning solutions in complex and dynamic combinatorial optimization domains, which introduces additional challenges. This paper introduces a novel QD framework for dynamic scheduling problems. We propose a map-building strategy that visualizes the solution space by linking heuristic genotypes to their behaviors, enabling their representation on a QD map. This map facilitates the discovery and maintenance of diverse scheduling heuristics. Additionally, we conduct experiments on both fixed and dynamically changing training instances to demonstrate how the map evolves and how the distribution of solutions unfolds over time. We also discuss potential future research directions that could enhance the learning process and broaden the applicability of QD algorithms to dynamic combinatorial optimization challenges.}, notes = {GECCO-2025 GP A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(catalano:2025:GECCO, author = {GianCarlo Antonino Pasquale Ignazio Catalano and Alexander Brownlee and David Cairns and Russell Ainslie and John McCall}, title = {Interpretable Decision Trees to Predict Solution Fitness}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1099--1107}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726387}, doi = {doi:10.1145/3712256.3726387}, size = {9 pages}, abstract = {Metaheuristic algorithms are powerful tools for tackling complex optimization problems, but their black-box nature often hinders user trust and understanding. This paper presents a novel methodology for enhancing the explainability of metaheuristics by employing decision trees with splitting criteria based on Partial Solutions. These represent beneficial sub-structures of solutions and provide insights into the problem landscape and solution characteristics. By constructing decision trees that consider the presence or absence of specific patterns in solutions, we produce a transparent model capable of predicting solution fitness.The proposed methodology is evaluated on a diverse set of benchmark problems and metaheuristic algorithms, demonstrating its effectiveness and flexibility as a post-hoc explainability tool.Our results show that our decision trees can match and usually surpass traditional methods in predicting the fitness of candidate solutions for the tested benchmark problems, with one of our methods demonstrating an improvement between 4.4\% and 16.7\% in R2 predictive performance for shallower trees trained on a Genetic Algorithm's data. These trees are able to maintain competitive predictive performance while using more interpretable splitting criteria.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(chen:2025:GECCO2, author = {Xiang-Ling Chen and Yi Mei and Mengjie Zhang}, title = {Learning Adaptive Neighborhood Search with Dual Operator Selection for Capacitated Vehicle Routing Problem}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1108--1116}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {adaptive operator selection, learn to optimize, meta-heuristic, capacitated vehicle routing problem, Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726372}, doi = {doi:10.1145/3712256.3726372}, size = {9 pages}, abstract = {The Capacitated Vehicle Routing Problem (CVRP) is a classic optimization problem with widespread applications in real-world scenarios. Existing research has developed numerous neighborhood operators capable of generating high-quality solutions. However, most neighborhood search algorithms either apply all operators in a predefined/random sequence or adaptively adjust only improvement operators, neglecting the learning of perturbation operators. This limitation often results in suboptimal performance. To address this issue, this paper proposes Learning Adaptive Neighborhood Search with Dual operator Selection (LANDS), an algorithm that integrates two RL-based controllers to adaptively select both improvement and perturbation operators simultaneously. Within this framework, the two controllers act as high-level guides, collaboratively directing the optimization process by selecting appropriate operators. Additionally, to reduce resource waste, a filtering mechanism is introduced to exclude operators deemed ineffective within the same improvement iteration, enhancing the algorithm's efficiency. The effectiveness of the proposed LANDS method is verified by a series of experiments.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(guo:2025:GECCO, author = {Hongshu Guo and Sijie Ma and Zechuan Huang and Yuzhi Hu and Zeyuan Ma and Xinglin Zhang and Yue-Jiao Gong}, title = {Reinforcement Learning-based Self-adaptive Differential Evolution through Automated Landscape Feature Learning}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1117--1126}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726309}, doi = {doi:10.1145/3712256.3726309}, size = {10 pages}, abstract = {Recently, Meta-Black-Box-Optimization (MetaBBO) methods significantly enhance the performance of traditional black-box optimizers through meta-learning flexible and generalizable meta-level policies that excel in dynamic algorithm configuration (DAC) tasks within the low-level optimization, reducing the expertise required to adapt optimizers for novel optimization tasks. Though promising, existing MetaBBO methods heavily rely on human-crafted feature extraction approach to secure learning effectiveness. To address this issue, this paper introduces a novel MetaBBO method that supports automated feature learning during the meta-learning process, termed as RLDE-AFL, which integrates a learnable feature extraction module into a reinforcement learning-based DE method to learn both the feature encoding and meta-level policy. Specifically, we design an attention-based neural network with mantissa-exponent based embedding to transform the solution populations and corresponding objective values during the low-level optimization into expressive landscape features. We further incorporate a comprehensive algorithm configuration space including diverse DE operators into a reinforcement learning-aided DAC paradigm to unleash the behavior diversity and performance of the proposed RLDE-AFL. Extensive benchmark results show that co-training the proposed feature learning module and DAC policy contributes to the superior optimization performance of RLDE-AFL to several advanced DE methods and recent MetaBBO baselines over both synthetic and realistic BBO scenarios.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(ma:2025:GECCO, author = {Zeyuan Ma and Hongqiao Lian and Wenjie Qiu and Yue-Jiao Gong}, title = {Accurate Peak Detection in Multimodal Optimization via Approximated Landscape Learning}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1127--1136}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726308}, doi = {doi:10.1145/3712256.3726308}, size = {10 pages}, abstract = {Detecting potential optimal peak areas and locating the accurate peaks in these areas are two major challenges in Multimodal Optimization problems (MMOPs). To address them, much efforts have been spent on developing novel searching operators, niching strategies and multi-objective problem transformation pipelines. Though promising, existing approaches more or less overlook the potential usage of landscape knowledge. In this paper, we propose a novel optimization framework tailored for MMOPs, termed as APDMMO, which facilitates peak detection via fully leveraging the landscape knowledge and hence capable of providing strong optimization performance on MMOPs. Specifically, we first design a novel surrogate landscape model which ensembles a group of non-linear activation units to improve the regression accuracy on diverse MMOPs. Then we propose a free-of-trial peak detection method which efficiently locates potential peak areas through back-propagation on the learned surrogate landscape model. Based on the detected peak areas, we employ SEP-CMAES for local search within these areas in parallel to further improve the accuracy of the found optima. Extensive benchmarking results demonstrate that APDMMO outperforms several up-to-date baselines. Further ablation studies verify the effectiveness of the proposed novel designs. The source-code is available at https://github.com/GMC-DRL/APDMMO.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(ma:2025:GECCO2, author = {Zeyuan Ma and Zhiyang Huang and Jiacheng Chen and Zhiguang Cao and Yue-Jiao Gong}, title = {Surrogate Learning in Meta-Black-Box Optimization: A Preliminary Study}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1137--1145}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726316}, doi = {doi:10.1145/3712256.3726316}, size = {9 pages}, abstract = {Recent Meta-Black-Box Optimization (MetaBBO) approaches have shown possibility of enhancing the optimization performance through learning meta-level policies to dynamically configure low-level optimizers. However, existing MetaBBO approaches potentially consume massive function evaluations to train their meta-level policies. Inspired by the recent trend of using surrogate models for cost-friendly evaluation of expensive optimization problems, in this paper, we propose a novel MetaBBO framework which combines surrogate learning process and reinforcement learning-aided Differential Evolution algorithm, namely Surr-RLDE, to address the intensive function evaluation in MetaBBO. Surr-RLDE comprises two learning stages: surrogate learning and policy learning. In surrogate learning, we train a Kolmogorov-Arnold Networks (KAN) with a novel relative-order-aware loss to accurately approximate the objective functions of the problem instances used for subsequent policy learning. In policy learning, we employ reinforcement learning (RL) to dynamically configure the mutation operator in DE. The learned surrogate model is integrated into the training of the RL-based policy to substitute for the original objective function, which effectively reduces consumed evaluations during policy learning. Extensive benchmark results demonstrate that Surr-RLDE not only shows competitive performance to recent baselines, but also shows compelling generalization for higher-dimensional problems. Further ablation studies underscore the effectiveness of each technical components in Surr-RLDE. We open-source Surr-RLDE at https://github.com/GMC-DRL/Surr-RLDE.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(mittermaier:2025:GECCO, author = {Michael Mittermaier and Takfarinas Saber and Goetz Botterweck}, title = {Learning Graph Configuration Spaces to Support Road Network Design Optimisation}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1146--1152}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726447}, doi = {doi:10.1145/3712256.3726447}, size = {7 pages}, abstract = {Genetic algorithms (GA) allow us to optimise graphs according to multiple objectives while considering many different constraints. These population-based algorithms assess the fitness of a high number of genomes. In the case of optimising road networks, a high number of fitness assessments leads to high computational costs of traffic simulations. In this work, we explore the application of learning graph configuration spaces to make efficient use of these traffic simulations by using learning model predictions for the majority of fitness assessments. In a controlled experiment, we compare the quality of GA optimisations with and without learning model predictions on the same simulation budget. Our results indicate that although we lose accuracy in the fitness assessments with predictions, the GA reliably finds road networks with better traffic flow and lower overall road length while using the same number of traffic simulations. We show that learning models can support GAs to make efficient use of the simulation budget and thus improve the optimisation. Future work is necessary to confirm these results for larger road networks.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(morales-paredes:2025:GECCO2, author = {Adrian Morales-Paredes and Julio Juarez and Jesus Falcon-Cardona and Hugo Terashima-Marin and Carlos {Coello Coello}}, title = {Automatic Design of Specialized Variation Operators for the Multi-Objective Quadratic Assignment Problem}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1153--1161}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, grammatical evolution, Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726456}, doi = {doi:10.1145/3712256.3726456}, size = {9 pages}, abstract = {The development of specialized, domain-specific operators has significantly enhanced the performance of evolutionary algorithms for solving optimization problems. However, creating such operators often requires substantial effort from human experts, making the process slow, resource-intensive, and heavily reliant on domain knowledge. To overcome these limitations, generation hyper-heuristics provide a framework for automating the design of variation operators by evolving combinations of heuristic components without direct expert input. In this work, we propose a generation hyper-heuristic method based on grammatical evolution to automatically design variation operators (crossover and mutation) tailored to the multi-objective quadratic assignment problem (mQAP)-a challenging combinatorial optimization problem with many real-world applications. Using the proposed method, variation operators were generated considering six mQAP instances with two and three objectives, leveraging MOEA/D as a multi-objective optimizer. For validation, the generated operators were evaluated on unseen instances. Our experimental results indicate that the evolved operators enhance the performance of MOEA/D compared to standard crossover operators. Furthermore, the top-performing operator in training did not always generalize best to larger instances, while some lower-ranked operators showed better adaptability. These results highlight the potential of automated operator design in effectively tackling complex optimization problems like the mQAP.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(nguyen:2025:GECCO2, author = {Tai Nguyen and Phong Le and Andre Biedenkapp and Carola Doerr and Nguyen Dang}, title = {On the Importance of Reward Design in Reinforcement Learning-based Dynamic Algorithm Configuration: A Case Study on {OneMax} with (1+(lambda,lambda))-GA}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1162--1171}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726395}, doi = {doi:10.1145/3712256.3726395}, size = {10 pages}, abstract = {Dynamic Algorithm Configuration (DAC) has garnered significant attention in recent years, particularly in the prevalence of machine learning and deep learning algorithms. Numerous studies have leveraged the robustness of decision-making in Reinforcement Learning (RL) to address the optimization challenges associated with algorithm configuration. However, making an RL agent work properly is a non-trivial task, especially in reward design, which necessitates a substantial amount of handcrafted knowledge based on domain expertise. In this work, we study the importance of reward design in the context of DAC via a case study on controlling the population size of the (1 + (lambda, lambda))-GA optimizing OneMax. We observed that a poorly designed reward can hinder the RL agent's ability to learn an optimal policy because of a lack of exploration, leading to both scalability and learning divergence issues. To address those challenges, we propose the application of a reward shaping mechanism to facilitate enhanced exploration of the environment by the RL agent. Our work not only demonstrates the ability of RL in dynamically configuring the (1 + (lambda, lambda))-GA, but also confirms the advantages of reward shaping in the scalability of RL agents across various sizes of OneMax problems.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(notice:2025:GECCO, author = {Danielle Notice and Hamed Soleimani and Nicos G. Pavlidis and Ahmed Kheiri and Mario Andres Munoz}, title = {Instance Space Analysis of the Capacitated Vehicle Routing Problem with Mixture Discriminant Analysis}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1172--1180}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {capacitated vehicle routing problem, instance space analysis, algorithm selection, supervised dimension reduction, discriminant analysis, Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726405}, doi = {doi:10.1145/3712256.3726405}, size = {9 pages}, abstract = {In this paper, we attempt a deeper understanding of the relative performance of two state-of-the-art metaheuristic solvers for the capacitated vehicle routing problem (CVRP). To this end, we employ a novel CVRP instance generator to expand the set of CVRP instances used to assess heuristics. This generator modifies existing problem instances using the outliers of node clusters to produce relevant new CVRP instances. We consider a large number of features to characterise each problem instance, and propose to use mixture discriminant analysis (MDA) to obtain both a low dimensional representation of the instance space and a classifier of algorithm performance. MDA has not been previously used in instance space analysis, and as a supervised dimension reduction method has the advantage that the tasks of dimension reduction and classification are handled in a unified framework (rather than two separate steps). The resulting predictive models perform as well as more complex classifiers that involve more tuning parameters and are computationally more expensive (like support vector machines). Our analysis highlights that the performance comparison between the two CVRP metaheuristics is nuanced and the best algorithm depends on the time budget, as well as certain key characteristics of the problem instance.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(petelin:2025:GECCO, author = {Gasper Petelin and Gjorgjina Cenikj}, title = {The Pitfalls of Benchmarking in Algorithm Selection: What We Are Getting Wrong}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1181--1189}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726336}, doi = {doi:10.1145/3712256.3726336}, size = {9 pages}, abstract = {Algorithm selection, aiming to identify the best algorithm for a given problem, plays a pivotal role in continuous black-box optimization. A common approach involves representing optimization functions using a set of features, which are then used to train a machine learning meta-model for selecting suitable algorithms. Various approaches have demonstrated the effectiveness of these algorithm selection meta-models. However, not all evaluation approaches are equally valid for assessing the performance of metamodels. We highlight methodological issues that frequently occur in the community and should be addressed when evaluating algorithm selection approaches. First, we identify flaws with the "leave-instance-out" evaluation technique. We show that non-informative features and meta-models can achieve high accuracy, which should not be the case with a well-designed evaluation framework. Second, we demonstrate that measuring the performance of optimization algorithms with metrics sensitive to the scale of the objective function requires careful consideration of how this impacts the construction of the meta-model, its predictions, and the model's error. Such metrics can falsely present overly optimistic performance assessments of the meta-models. This paper emphasizes the importance of careful evaluation, as loosely defined methodologies can mislead researchers, divert efforts, and introduce noise into the field.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(schede:2025:GECCO, author = {Elias Schede and Moritz Seiler and Kevin Tierney and Heike Trautmann}, title = {Deep reinforcement learning for instance-specific algorithm configuration}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1190--1198}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {algorithm configuration, parameter tuning, reinforcement learning, Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726480}, doi = {doi:10.1145/3712256.3726480}, size = {9 pages}, abstract = {Optimization algorithms contain parameters that greatly influence their behavior. Finding the right settings for parameters through automated algorithm configuration has become a critical component of designing competitive algorithms. While traditional offline configurators tackle this problem by finding one configuration that works well for a set of instances, instance-specific algorithm configuration uses features of the instances to provide configurations that are tailored to each instance to maximize performance. We provide the first instance-specific algorithm configurator based on deep reinforcement learning that can be used in general algorithm configuration settings. Our method is able to handle large, mixed discrete and continuous search spaces and only requires a small number of instances for training. We can show that our configurator provides improvements over the state-of-the-art instance-specific configurators ISAC and Hydra on a wide range of problem domains.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(schaepermeier:2025:GECCO, author = {Lennart Schaepermeier}, title = {Greedy Restart Schedules: A Baseline for Dynamic Algorithm Selection on Numerical Black-box Optimization Problems}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1199--1207}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {heuristic optimization, continuous optimization, algorithm scheduling, dynamic algorithm selection, benchmarking, performance analysis, Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726408}, doi = {doi:10.1145/3712256.3726408}, size = {9 pages}, abstract = {In many optimization domains, there are multiple different solvers that contribute to the overall state-of-the-art, each performing better on some, and worse on other types of problem instances. Metaalgorithmic approaches, such as instance-based algorithm selection, configuration and scheduling, aim to close this gap by extracting the most performance possible from a set of (configurable) optimizers. In this context, the best performing individual algorithms are often hand-crafted hybrid heuristics which perform many restarts of fast local optimization approaches. However, data-driven techniques to create optimized restart schedules have not yet been extensively studied.Here, we present a simple scheduling approach that iteratively selects the algorithm performing best on the distribution of unsolved training problems at time of selection, resulting in a problem-independent solver schedule. We demonstrate our approach using well-known optimizers from numerical black-box optimization on the BBOB testbed, bridging much of the gap between single and virtual best solver from the original portfolio across various evaluation protocols. Our greedy restart schedule presents a powerful baseline for more complex dynamic algorithm selection models.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(wang:2025:GECCO, author = {Ruilin Wang and Xiang Feng and Huiqun Yu and Edmund Lai}, title = {Residual Learning Inspired Crossover Operator and Strategy Enhancements for Evolutionary Multitasking}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Marie-Eleonore Kessaci and Anna V. Kononova}, pages = {1208--1216}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Learning for Evolutionary Computation}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726322}, doi = {doi:10.1145/3712256.3726322}, size = {9 pages}, abstract = {In evolutionary multitasking, strategies such as crossover operators and skill factor assignment are critical for effective knowledge transfer. Existing improvements to crossover operators primarily focus on low-dimensional variable combinations, such as arithmetic crossover or partially mapped crossover, which are insufficient for modeling complex high-dimensional interactions. Moreover, static or semi-dynamic crossover strategies fail to adapt to the dynamic dependencies among tasks. In addition, current Multifactorial Evolutionary Algorithm frameworks often rely on fixed skill factor assignment strategies, lacking flexibility. To address these limitations, this paper proposes the Multifactorial Evolutionary AlgorithmResidual Learning (MFEA-RL) method based on residual learning. The method employs a Very Deep Super-Resolution (VDSR) model to generate high-dimensional residual representations of individuals, enhancing the modeling of complex relationships within dimensions. A ResNet-based mechanism dynamically assigns skill factors to improve task adaptability, while a random mapping mechanism efficiently performs crossover operations and mitigates the risk of negative transfer. Theoretical analysis and experimental results show that MFEA-RL outperforms state-of-the-art multitasking algorithms. It excels in both convergence and adaptability on standard evolutionary multitasking benchmarks, including CEC2017-MTSO and WCCI2020-MTSO. Additionally, its effectiveness is validated through a real-world application scenario.}, notes = {GECCO-2025 L4EC A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(abrantes:2025:GECCO, author = {Joao Abrantes and Robert Lange and Yujin Tang}, title = {Competition and Attraction Improve Model Fusion}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Bing Xue and Dennis Wilson}, pages = {1217--1225}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Neuroevolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726329}, doi = {doi:10.1145/3712256.3726329}, size = {9 pages}, abstract = {Model merging is a powerful technique for integrating the specialized knowledge of multiple machine learning models into a single model. However, existing methods require manually partitioning model parameters into fixed groups for merging, which restricts the exploration of potential combinations and limits performance. To overcome these limitations, we propose Model Merging of Natural Niches (M2N2), an evolutionary algorithm with three key features: (1) dynamic adjustment of merging boundaries to progressively explore a broader range of parameter combinations; (2) a diversity preservation mechanism inspired by the competition for resources in nature, to maintain a population of diverse, high-performing models that are particularly well-suited for merging; and (3) a heuristic-based attraction metric to identify the most promising pairs of models for fusion. Our experimental results demonstrate, for the first time, that model merging can be used to evolve models entirely from scratch. Specifically, we apply M2N2 to evolve MNIST classifiers from scratch and achieve performance comparable to CMA-ES, while being computationally more efficient. Furthermore, M2N2 scales to merge specialized language and image generation models, achieving state-of-the-art performance. Notably, it preserves crucial model capabilities beyond those explicitly optimized by the fitness function, highlighting its robustness and versatility. Our code is available at https://github.com/AnonScientist/natural_niches.}, notes = {GECCO-2025 NE A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(chen:2025:GECCO3, author = {Chih-Ling Chen and Kai-Chiang Wu and Ning-Chi Huang}, title = {Integrating Neural Architecture Search and Rematerialization for Efficient On-Device Learning}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Bing Xue and Dennis Wilson}, pages = {1226--1234}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {on-device learning, efficient training, neural architecture search, rematerialization, Neuroevolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726482}, doi = {doi:10.1145/3712256.3726482}, size = {9 pages}, abstract = {Deep neural networks (DNNs) have notable performance in many fields, such as computer vision. Training a neural network on an edge device, commonly called on-device learning, has grown crucial for applications demanding real-time processing and enhanced privacy. However, existing on-device learning methods often face limitations, such as decreasing application accuracy, causing complexity in design and implementation, and increasing computational overhead, all of which hinder their effectiveness in reducing memory usage. In this paper, we address the issue by inspecting the memory usage of training a DNN, analyzing the effects of different on-device learning strategies, and introducing a framework that integrates neural architecture search (NAS) and rematerialization. The supernet of NAS can provide a population of compressed subnets/architectures to be trained without additional computational overhead, while rematerialization can mitigate memory consumption without accuracy loss. By leveraging the memory-saving effect of both supernet-based model compression and rematerialization, our proposed method can obtain suitable models that fit within the memory constraint while achieving a better trade-off between training time and model performance. In the experiments, we used complex datasets (CIFAR-100 and CUB-200) to fine-tune models on Raspberry Pi. The experimental results represent the effectiveness of our method in real-world on-device learning scenarios.}, notes = {GECCO-2025 NE A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(feiden:2025:GECCO, author = {Arno Feiden and Jochen Garcke}, title = {Diversity in Reinforcement Learning Through the Occupancy Measure}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Bing Xue and Dennis Wilson}, pages = {1235--1245}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Neuroevolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726337}, doi = {doi:10.1145/3712256.3726337}, size = {11 pages}, abstract = {Quality-Diversity algorithms search for a set of diverse, high-performing solutions to optimization problems, including reinforcement learning problems. In the case of reinforcement learning problems, Quality-Diversity algorithms foster diversity by differentiating solutions using behaviour descriptors. We introduce a straightforward, powerful approach to generically characterise behaviour using the so-called occupancy measure. Our approach avoids the manual definition of behaviour descriptors and does not rely on further black-box learning.We investigate four established benchmark problems inspired by robotics, concerning locomotion and maze navigation. To measure the ability to overcome local optima we consider the number of solved configurations and the maximum average score. The use of the occupancy measure is competitive with problem-specific, custom behaviour descriptors and superior to an established generic behaviour descriptor. Our work contributes to the establishment of MAP-Elites as a versatile, robust, out-of-the-box solver for complex non-convex reinforcement learning scenarios.}, notes = {GECCO-2025 NE A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(huang:2025:GECCO, author = {Junhao Huang and Bing Xue and Yanan Sun and Mengjie Zhang}, title = {Evolving Comprehensive Proxies for Zero-Shot Neural Architecture Search}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Bing Xue and Dennis Wilson}, pages = {1246--1254}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Neuroevolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726315}, doi = {doi:10.1145/3712256.3726315}, size = {9 pages}, abstract = {Neural architecture search (NAS) has emerged as a promising technology for automatically designing deep neural network (DNN) architectures. However, its development is significantly constrained by the prohibitively high computational cost of architecture evaluations. Recently, zero-shot NAS has addressed this challenge by employing zero-cost proxies to evaluate candidate architectures without expensive gradient training, effectively mitigating the timeintensive nature of NAS. However, a major limitation is that most existing zero-cost proxies focus narrowly on a single aspect of DNNs, resulting in biased evaluations with generally weak correlations to the network's performance. In this work, we address this issue by assembling four distinct zero-cost proxies in a nonlinear fashion to provide a comprehensive evaluation of DNNs across multiple dimensions, including expressivity, convergence, generalization, and parameter saliency. Furthermore, we develop an adaptive particle swarm optimization-based approach to effectively evolve the coefficients of each base proxy in the ensemble for task-specific optimization. Extensive experimental results on various NAS benchmarks and open-domain search spaces demonstrate the effectiveness of the proposed method. Our findings show that the complementarity of zero-cost proxies greatly improves the reliability of performance evaluation, thereby enabling zero-shot NAS to identify more promising network architectures.}, notes = {GECCO-2025 NE A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(montoya:2025:GECCO, author = {Oleg Montoya and Frantisek Srb and Djordje Grbic and Sebastian Risi}, title = {{CPPN2WFC:} Extending Wave Function Collapse to Generate Globally Coherent Content}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Bing Xue and Dennis Wilson}, pages = {1255--1263}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {neuroevolution, CPPN, NEAT, generative art, PCG, WFC}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726401}, doi = {doi:10.1145/3712256.3726401}, size = {9 pages}, abstract = {Procedural content generation (PCG) enables the creation of vast, varied, and aesthetically rich environments with minimal manual effort. One of the most widely used techniques for procedural map generation is Wave Function Collapse (WFC), a constraint-based algorithm that synthesizes game maps by propagating local patterns while ensuring global consistency. However, despite its effectiveness, WFC often produces repetitive structures and lacks the ability to introduce higher-order spatial coherence or emergent design patterns. This paper explores whether combining Compositional Pattern Producing Networks (CPPN) and WFC - a hybrid method we term CPPN2WFC - leads to more structured and visually compelling game maps compared to using WFC or CPPNs alone. CPPNs, which are artificial neural networks with a selection of different activation functions, have been shown to generate intricate patterns and organic-like structures when evolved through NEAT, a method that dynamically evolves both network topology and weights over generations. By integrating CPPNs into the WFC framework, we introduce an additional layer of flexibility, allowing both constraint satisfaction and high-level structural control. We conduct comparative experiments through an Interactive Evolutionary interface and user study. Main results show that compared to CPPNs or WFC alone, CPPN2WFC strikes a balance between producing global and local patterns.}, notes = {GECCO-2025 NE A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(kubalik:2025:GECCO, author = {Jiri Kubalik and Robert Babuska}, title = {Neuro-Evolutionary Approach to Physics-Aware Symbolic Regression}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Bing Xue and Dennis Wilson}, pages = {1264--1272}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, Neuroevolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726434}, doi = {doi:10.1145/3712256.3726434}, size = {9 pages}, abstract = {Symbolic regression is a technique that can automatically derive analytic models from data. Traditionally, symbolic regression has been implemented primarily through genetic programming that evolves populations of candidate solutions sampled by genetic operators, crossover and mutation. More recently, neural networks have been employed to learn the entire analytical model, i.e., its structure and coefficients, using regularized gradient-based optimization. Although this approach tunes the model's coefficients better, it is prone to premature convergence to suboptimal model structures. Here, we propose a neuro-evolutionary symbolic regression method that combines the strengths of evolutionary-based search for optimal neural network (NN) topologies with gradient-based tuning of the network's parameters. Due to the inherent high computational demand of evolutionary algorithms, it is not feasible to learn the parameters of every candidate NN topology to the full convergence. Thus, our method employs a memory-based strategy and population perturbations to enhance exploitation and reduce the risk of being trapped in suboptimal NNs. In this way, each NN topology can be trained using only a short sequence of back-propagation iterations. The proposed method was experimentally evaluated on three real-world test problems and has been shown to outperform other NN-based approaches regarding the quality of the models obtained.}, notes = {GECCO-2025 NE A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(long:2025:GECCO, author = {Quan Long and Bin Wang and Bing Xue and Mengjie Zhang}, title = {A Multi-Objective Approach to Optimizing Kolmogorov-Arnold Networks for Classification Tasks to Balance Accuracy and Interpretability}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Bing Xue and Dennis Wilson}, pages = {1273--1281}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {kolmogorov-arnold networks, multi-objective optimization, NSGA-II, interpretability, Neuroevolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726354}, doi = {doi:10.1145/3712256.3726354}, size = {9 pages}, abstract = {Kolmogorov-Arnold Network (KAN) has recently gained attention for its strong performance in both accuracy and interoperability. However, a trade-off often exists between these two objectives. To address this challenge, we propose MO-KAN, a novel multi-objective optimization framework based on the NSGA-II algorithm. MO-KAN optimizes the trade-off between accuracy and interpretability by leveraging NSGA-II to generate Pareto-optimal solutions, from which the most suitable solution is selected. Additionally, MO-KAN introduces a novel interpretability metric that quantifies the interpretability of KAN models by evaluating their simplicity, based on connection count and structural complexity, with the assumption that simpler models are more interpretable. Experimental results validate MO-KAN's effectiveness on four toy datasets from the original KAN study, achieving optimal structures while maintaining interpretability. Furthermore, experiments on five classification datasets from the UCI repository demonstrate that MO-KAN outperforms existing methods in achieving a better trade-off. By slightly compromising accuracy, MO-KAN identifies significantly more interpretable solutions, highlighting its potential for applications where interpretability is critical.}, notes = {GECCO-2025 NE A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(mitsides:2025:GECCO, author = {Konstantinos Mitsides and Maxence Faldor and Antoine Cully}, title = {Scaling Policy Gradient Quality-Diversity with Massive Parallelization via Behavioral Variations}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Bing Xue and Dennis Wilson}, pages = {1282--1290}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Neuroevolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726335}, doi = {doi:10.1145/3712256.3726335}, size = {9 pages}, abstract = {Quality-Diversity optimization comprises a family of evolutionary algorithms aimed at generating a collection of diverse and high-performing solutions. MAP-Elites (ME), a notable example, is used effectively in fields like evolutionary robotics. However, the reliance of ME on random mutations from Genetic Algorithms limits its ability to evolve high-dimensional solutions. Methods proposed to overcome this include using gradient-based operators like policy gradients or natural evolution strategies. While successful at scaling ME for neuroevolution, these methods often suffer from slow training speeds, or difficulties in scaling with massive parallelization due to high computational demands or reliance on centralized actor-critic training. In this work, we introduce a fast, sample-efficient ME based algorithm capable of scaling with massive parallelization, significantly reducing runtimes without compromising performance. Our method, ASCII-ME, unlike existing policy gradient quality-diversity methods, does not rely on centralized actor-critic training. It performs behavioral variations based on time step performance metrics and maps these variations to solutions using policy gradients. Our experiments show that ASCII-ME can generate a diverse collection of high-performing deep neural network policies in less than 250 seconds on a single GPU. Additionally, it operates on average, five times faster than state-of-the-art algorithms while maintaining competitive sample efficiency.}, notes = {GECCO-2025 NE A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(patterson:2025:GECCO, author = {Evan Patterson and Joshua Karns and Zimeng Lyu and Travis Desell}, title = {Visualizing the Dynamics of Neuroevolution with Genetic Distance Projections}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Bing Xue and Dennis Wilson}, pages = {1291--1299}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Neuroevolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726457}, doi = {doi:10.1145/3712256.3726457}, size = {9 pages}, abstract = {Evolutionary algorithms have shown substantial progress in recent years, especially in neural architecture search applications, or neuroevolution. Despite their effectiveness, analyzing and understanding the evolutionary paths these algorithms traverse to reach solutions remains challenging. Often these algorithms involve distributed computing strategies, which can include subpopulations or islands, and they explore massive or even unbounded search spaces which can include both weights and architecture, in both continuous and non-continuous domains. Manually examining individual solutions to understand the evolutionary dynamics is often infeasible due to large population sizes, large genome sizes, and high generation counts. This work introduces a new methodology for visualizing neuroevolution population dynamics called genetic distance projections, along with a novel neural network based method for generating these representations. We evaluate this methodology empirically and find it performs better than other traditional methods in generating these representations. We further validate the usefulness of these visualizations using case studies from EXAMM, a long standing neuroevolution algorithm, in which one case study even led to finding and fixing a bug in EXAMM's algorithm.}, notes = {GECCO-2025 NE A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(pinto:2025:GECCO, author = {Rafael Pinto and Anderson Tavares}, title = {Neuroevolution of Self-Attention Over Proto-Objects}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Bing Xue and Dennis Wilson}, pages = {1300--1308}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Neuroevolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726451}, doi = {doi:10.1145/3712256.3726451}, size = {9 pages}, abstract = {Proto-objects - image regions that share common visual properties - offer a promising alternative to traditional attention mechanisms based on rectangular-shaped image patches in neural networks. Although previous work demonstrated that evolving a patch-based hard-attention module alongside a controller network could achieve state-of-the-art performance in visual reinforcement learning tasks, our approach leverages image segmentation to work with higher-level features. By operating on proto-objects rather than fixed patches, we significantly reduce the representational complexity: each image decomposes into fewer proto-objects than regular patches, and each proto-object can be efficiently encoded as a compact feature vector. This enables a substantially smaller self-attention module that processes richer semantic information. Our experiments demonstrate that this proto-object-based approach matches or exceeds the state-of-the-art performance of patch-based implementations with 62\% less parameters and 2.6 times less training time.}, notes = {GECCO-2025 NE A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(zhou:2025:GECCO, author = {Yuyang Zhou and Ferrante Neri and Yew-Soon Ong and Ruibin Bai}, title = {{SiamNAS:} Siamese Surrogate Model for Dominance Relation Prediction in Multi-objective Neural Architecture Search}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Bing Xue and Dennis Wilson}, pages = {1309--1318}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Neuroevolution}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726359}, doi = {doi:10.1145/3712256.3726359}, size = {10 pages}, abstract = {Modern neural architecture search (NAS) is inherently multi-objective balancing trade-offs such as accuracy, parameter count, and computational cost. This complexity makes NAS computationally expensive and nearly impossible to solve without efficient approximations. To address this, we propose a novel surrogate modelling approach that leverages an ensemble of Siamese network blocks to predict dominance relationships between candidate architectures. Lightweight and easy to train, the surrogate achieves 92\% accuracy and replaces the crowding distance calculation in the survivor selection strategy with a heuristic rule based on model size. Integrated into a framework termed SiamNAS, this design eliminates costly evaluations during the search process. Experiments on NAS-Bench-201 demonstrate the framework's ability to identify Pareto-optimal solutions with significantly reduced computational costs. The proposed SiamNAS identified a final non-dominated set containing the best architecture in NAS-Bench-201 for CIFAR-10 and the second-best for ImageNet, in terms of test error rate, within 0.01 GPU days. This proof-of-concept study highlights the potential of the proposed Siamese network surrogate model to generalise to multi-tasking optimisation, enabling simultaneous optimisation across tasks. Additionally, it offers opportunities to extend the approach for generating Sets of Pareto Sets (SOS), providing diverse Pareto-optimal solutions for heterogeneous task settings.}, notes = {GECCO-2025 NE A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(barbosa:2025:GECCO, author = {Guilherme Barbosa and Pedro Pereira and Vasco Abelha and Rui Mendes and Paulo Cortez}, title = {A Dijkstra Seeded Evolutionary Multiobjective Optimization System for a Sustainable User Multimodal Transport Routing}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1319--1327}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {evolutionary multi-objective optimization, mobility as a service, Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726363}, doi = {doi:10.1145/3712256.3726363}, size = {9 pages}, abstract = {Assuming a Mobility as a Service (MaaS) concept, this paper presents a sustainable user multimodal public transport query system that minimizes both the travel time and carbon footprint (CO2). In particular, we propose a novel Dijkstra (DJ) Seeded Non-dominated Sorting Genetic Algorithm II (NSGA-II) that is termed DS-NSGA-II. Our DS-NSGA-II method adopts multigraph data and a flexible integer solution representation that can be applied to any metropolitan area and set of public transports (including the walking option). Also, it assumes real-world transport time schedules and other realistic estimates (e.g., transport transit times). As a use case, we explore real-world data from the Oporto city, corresponding to a large multigraph with about 2,500 geographic nodes and 287,000 transportation edges regarding bus, metro and walking connections. Several experiments were held, assuming three types of realistic routing query categories (easy, medium, hard). Overall, competitive results were obtained by the proposed DS-NSGA-II method when compared with a standard NSGA-II and DJ approaches.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(bentley:2025:GECCO, author = {Peter Bentley and Soo Ling Lim and Fuyuki Ishikawa}, title = {{CLEAR:} Cue Learning using Evolution for Accurate Recognition Applied to Sustainability Data Extraction}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1328--1336}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726317}, doi = {doi:10.1145/3712256.3726317}, size = {9 pages}, abstract = {Large Language Model (LLM) image recognition is a powerful tool for extracting data from images, but accuracy depends on providing sufficient cues in the prompt - requiring a domain expert for specialized tasks. We introduce Cue Learning using Evolution for Accurate Recognition (CLEAR), which uses a combination of LLMs and evolutionary computation to generate and optimize cues such that recognition of specialized features in images is improved. It achieves this by auto-generating a novel domain-specific representation and then using it to optimize suitable textual cues with a genetic algorithm. We apply CLEAR to the real-world task of identifying sustainability data from interior and exterior images of buildings. We investigate the effects of using a variable-length representation compared to fixed-length and show how LLM consistency can be improved by refactoring from categorical to real-valued estimates. We show that CLEAR enables higher accuracy compared to expert human recognition and human-authored prompts in every task with error rates improved by up to two orders of magnitude and an ablation study evincing solution concision.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(chen:2025:GECCO4, author = {Tai-You Chen and Feng-Feng Wei and Wei-Neng Chen}, title = {Multi-Agent Swarm Optimization for Decentralized Energy Management Considering Game Behaviors of Electric Vehicles}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1337--1344}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {distributed optimization, particle swarm optimization, multi-agent systems, power dispatch, energy management, electric vehicles, Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726460}, doi = {doi:10.1145/3712256.3726460}, size = {8 pages}, abstract = {With the rapid growth of electric vehicles (EVs) in modern cities, it is worth studying the distributed energy management with EV charging (DER-EV). Due to the autonomous decision making behaviors of EVs, it is challenging to optimize the global objective and balance the supply and demand without a central node. In this work, we propose a multi-agent particle swarm optimization algorithm (MASOIE-G) for the DER-EV problem. In MASOIE-G, each agent maintains a particle swarm, and each particle in the swarm represents a candidate energy management solution. Through peer-to-peer communication, agents collaboratively evolve their particle swarms and optimize the global objective. To balance the supply and demand for the power grid with the participation of EVs, we develop an evolutionary game strategy for adaptive pricing at the phase of internal learning. To make agents cooperate to optimize the global objective with limited data and communication range, we design an external learning method with variable neighbor weights to help agents learn effective knowledge from multiple neighbors. Experimental validation on the IEEE 39-bus system shows that our algorithm can effectively respond to various conditions of supply and demand, achieving better performance than existing distributed evolutionary algorithms.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(chitty:2025:GECCO, author = {Darren Chitty and Ayah Helal and Sareh Rowlands and Craig Willis and Christopher Underwood and Ed Keedwell}, title = {Ensemble Phased Genetic Programming for Roundabout Turn Restriction Prediction}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1345--1353}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726424}, doi = {doi:10.1145/3712256.3726424}, size = {9 pages}, abstract = {Ensemble methods are among the best performing in the machine learning literature, often outperforming single methods in training accuracy and the prevention of overfitting. This work builds on the previously successful phased genetic programming (GP) approach to build ensembles of GP trees to create ensemble phased GP (EPGP). The method is tested in a real-world transportation modelling problem, the roundabout (traffic circle, rotary) turn restriction problem using data from OpenStreetMap, an important and time-consuming element of the traffic modelling process. EPGP is compared with standard and phased GP formulations and representative algorithms from the machine learning literature and is found to outperform them on this task.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(gaspar-cunha:2025:GECCO, author = {Antonio Gaspar-Cunha and Joao Melo and Tomas Marques and Antonio Pontes}, title = {Optimization of Conformal Cooling Channels for Injection Molding Multi-Objective Artificial Intelligence Techniques}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1354--1361}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726389}, doi = {doi:10.1145/3712256.3726389}, size = {8 pages}, abstract = {The injection molding process is widely used for manufacturing plastic components, where proper mold design is crucial to minimize defects. This study explores the optimization of conformal cooling channels (CCC) using artificial intelligence (AI) techniques, including principal component analysis (PCA), multi-objective evolutionary algorithms (MOEA), and artificial neural networks (ANN), integrated with numerical simulations. The methodology enhanced thermal efficiency and reduced cycle time in a cylindrical part with complex cooling requirements. Results demonstrated significant improvements in temperature uniformity and defect reduction, underscoring the potential of AI-driven optimization in advanced mold design. Furthermore, this study addresses a critical gap in existing optimization methodologies: selecting objectives. Leveraging Nonlinear Principal Component Analysis (NL-PCA), the approach identifies and prioritizes key objectives, enabling a focused and computationally efficient optimization process. Additionally, the study highlights the integration of data-driven approaches to streamline decision-making, ensuring a balance between computational efficiency and practical feasibility. The presented approach also serves as a stepping stone for integrating sustainability metrics, making it a critical contribution to the evolution of manufacturing technologies.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(grouwels:2025:GECCO, author = {Joris Grouwels and Nicolas Jonason and Bob L. T. Sturm}, title = {Exploring the Expressive Space of an Articulatory Vocal Modal using Quality-Diversity Optimization with Multimodal Embeddings}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1362--1370}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726313}, doi = {doi:10.1145/3712256.3726313}, size = {9 pages}, abstract = {Knowing which sounds can be produced by a simulated vocal model and how they are connected to its articulatory behavior is not trivial. Being able to map this out can be interesting for applications that make use of the extended capabilities of a voice, e.g., singing or vocal imitations. We present a method that achieves this for a state-of-the-art articulatory vocal model (VocalTractLab) by combining it with a recent Quality-Diversity algorithm (CMA-MAE) and audio embeddings obtained through a multi-modal pretrained model (CLAP). The text-capabilities of CLAP make it possible to steer the exploration through a text prompt. We show that the method explores more efficiently than a random sampling baseline, covering more of the measure space and achieving higher objective scores. We provide several listening examples and the source code for a scalable implementation.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(jiao:2025:GECCO, author = {Jian Jiao and Liu Yuan}, title = {{GA-PRE:} A Genetic Algorithm-Based Automatic Data Preprocessing Algorithm}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1371--1378}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726312}, doi = {doi:10.1145/3712256.3726312}, size = {8 pages}, abstract = {In the fields of data mining and machine learning, data preprocessing is a critical step in improving model performance. Traditional preprocessing methods are based on expert experience and manual adjustment. Although experienced data analysts can effectively preprocess data, individual experiences are difficult to quantify, leading to inconsistent data quality and a time-consuming process. To address this issue, this study proposes an automatic data preprocessing algorithm based on genetic algorithms (GA-PRE). The algorithm optimizes the accuracy of machine learning models as a fitness metric, exploring the search space of preprocessing steps to automatically select the most suitable combination of preprocessing methods. Automatically performs preprocessing tasks including data imputation, feature selection, duplicate value handling, and standardization. The algorithm has been extensively tested on multiple public datasets, and the experimental results demonstrate that, compared to three other preprocessing methods, the genetic algorithm-based preprocessing algorithm can significantly improve the accuracy of the model and the quality of the dataset.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(khan:2025:GECCO, author = {Muhammad Wishal Khan and Hooman {Oroojeni M. J.} and Bal Sanghera and Tim Blackwell and Mohammad Majid al-Rifaie}, title = {Tomographic Reconstruction with Real-time a priori Acquisition}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1379--1387}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {swarm intelligence, tomographic reconstruction, a priori, denoising, regularization, Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726365}, doi = {doi:10.1145/3712256.3726365}, size = {9 pages}, abstract = {An a priori construction technique is proposed, using a minimal swarm optimizer to achieve enhanced reconstruction accuracy. Central to the method is the incorporation of a priori information, constructed dynamically without prior knowledge of material properties, positions, or structural details. Instead, this a priori information is derived directly from sinogram projections for each defined angle in the set. A novel dynamic masking strategy leverages these sinogram-derived values to identify certain entries, generating a priori data structure that removes regions in the reconstructed image corresponding to zero sinogram values, thereby achieving theoretical dimensionality reduction. This approach effectively eliminates noise and artifacts, resulting in significantly lower reconstruction and reproduction errors.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(llamazares-lopez:2025:GECCO, author = {Marcos {Llamazares Lopez} and Daniel Parra and Jose Manuel {Velasco Cabo} and Oscar Garnica and Rafael Jacinto {Villanueva Mico} and J. Ignacio Hidalgo}, title = {Unveiling the dynamics of {NOx} pollution in internal combustion engines by Structured Grammatical Evolution}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1388--1396}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, grammatical evolution, Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726421}, doi = {doi:10.1145/3712256.3726421}, size = {9 pages}, abstract = {The formation of nitrogen oxides (NOx) in combustion systems is notable for its harmful impact on public health and the environment. Therefore, it is imperative to develop models to predict NOx formation in different situations. These models are designed to capture the characteristics of three distinct engine operating states: nominal, startup, and saturation. The nominal state represents the typical operating conditions, the startup state refers to the initial phase of the operation of the engine, and the saturation state corresponds to the operation of the engine at its maximum capacity. We applied dynamic structured grammatical evolution to obtain a set of interpretable expressions, which are mathematical representations capable of capturing the dynamics of NOx formation in combustion systems and that can be easily interpreted. These models were compared with traditional differential equation-based models to assess their interpretability and predictive accuracy for the three scenarios. Through our approach, we obtained a set of interpretable expressions that improved those obtained by a differential equation-based mathematical model, providing a more transparent and intuitive understanding of the system's behavior. Our technique seeks to unveil the dynamics of NOx formation processes that could significantly reduce NOx emissions and mitigate their impact on global environmental pollution.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(lunet:2025:GECCO, author = {Miguel Lunet and Daniela Fernandes and Fabio Neves-Moreira and Pedro Amorim}, title = {Symbolic Pricing Policies for Attended Home Delivery - the Case of an Online Retailer}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1397--1405}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, attended home delivery, dynamic pricing, sequential decision-making, vehicle routing, explainable artificial intelligence, Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726361}, doi = {doi:10.1145/3712256.3726361}, size = {9 pages}, abstract = {To get products delivered, clients and retailers agree on a delivery time window. We collaborated with an online retailer to develop a real-world application aimed at dynamically determining the delivery fee for each time window while ensuring the explainability of the pricing policy. This sequential decision-making problem arises as new customers continuously arrive. The objective is to maximize the final profit, given by the sum of baskets and delivery fees, discounted by the transportation and fleet costs. As multiple customers share the same delivery route, the costs are distributed among them, complicating the calculation of the marginal cost of each customer. Our study employs Genetic Programming (GP) to create explainable and easy-to-compute pricing policies to determine the delivery fees. These policies, expressed as mathematical formulas, rank price panels - combinations of time slots and corresponding fees - to identify optimal prices for each customer. The inputs to the GP algorithm capture the current state of the system, including factors such as capacity, customer location, and basket value. The resulting expressions offer operational managers a transparent pricing policy that allows them to maximize total profit.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(maci\k{a}zek:2025:GECCO, author = {Jakub Maci\k{a}zek and Michal Przewozniczek and Jonas Schwaab}, title = {Seeking and leveraging alternative variable dependency concepts in gray-box-elusive bimodal land-use allocation problems}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1406--1414}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {land-use allocation, multi-objective optimization, variable dependency, gray-box optimization, genetic algorithms, evolutionary algorithms, optimization, Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726419}, doi = {doi:10.1145/3712256.3726419}, size = {9 pages}, abstract = {Solving land-use allocation problems can help us to deal with some of the most urgent global environmental issues. Since these problems are NP-hard, effective optimizers are needed to handle them. The knowledge about variable dependencies allows for proposing such tools. However, in this work, we consider a real-world multi-objective problem for which standard variable dependency discovery techniques are inapplicable. Therefore, using linkage-based variation operators is unreachable. To address this issue, we propose a definition of problem-dedicated variable dependency. On this base, we propose obtaining masks of dependent variables. Using them, we construct three novel crossover operators. The results concerning real-world test cases show that introducing our propositions into two well-known optimizers (NSGA-II, MOEA/D) dedicated to multi-objective optimization significantly improves their effectiveness.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(markovic:2025:GECCO, author = {Tijana Markovic and Pontus Lidholm and Per Erik Strandberg and Miguel Leon}, title = {Feature Selection Using Genetic Algorithm for Intrusion Detection on Resource-Constrained Edge Devices}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1415--1423}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {intrusion detection, machine learning, genetic algorithm, feature selection, edge computing, embedded system, Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726347}, doi = {doi:10.1145/3712256.3726347}, size = {9 pages}, abstract = {Intrusion Detection (ID) systems play a crucial role in protecting computer networks from growing number of cyber threats, with Machine Learning (ML) algorithms emerging as highly effective tools in strengthening ID performance. In recent years, there has been a notable shift towards deploying ML algorithms for ID directly on edge devices, to enhance performance and increase data privacy. However, this requires ML models to be optimized for resource-constrained devices. This paper is focused on applying genetic algorithm for feature selection in ML-based ID systems deployed on edge devices. It investigates how feature selection impacts the performance of various ML algorithms, including decision tree, random forest, and artificial neural network. The study is conducted using publicly available Westermo network traffic dataset and evaluated for live network traffic classification on an edge device manufactured by Westermo Network Technologies. Using only features selected by genetic algorithm resulted in a reduction of 14--26\% for peak memory consumption and 23--40\% for total memory consumption and decreased detection time by 24--69\%, depending on the algorithm, while maintaining system classification performance. Together with the increasing computational power of edge devices, these results facilitate the application of edge ML by reducing system requirements concerning memory and processing time.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(millar:2025:GECCO, author = {Robert Millar and Jinglai Li}, title = {Bayesian Optimization for {CVaR-based} portfolio optimization}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1424--1432}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726307}, doi = {doi:10.1145/3712256.3726307}, size = {9 pages}, abstract = {Optimal portfolio allocation is often formulated as a constrained risk problem, where one aims to minimize a risk measure subject to some performance constraints. This paper presents new Bayesian Optimization algorithms for such constrained minimization problems, seeking to minimize the conditional value-at-risk (a computationally intensive risk measure) under a minimum expected return constraint. The proposed algorithms use a new acquisition function, which drives sampling towards the optimal region. Additionally, a new two-stage procedure is developed, which significantly reduces the number of evaluations of the expensive-to-evaluate objective function. The proposed algorithm's competitive performance is demonstrated through practical examples.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(megane:2025:GECCO, author = {Jessica Megane and Nuno Lourenco and J. Ignacio Hidalgo and Penousal Machado}, title = {Contribution of Probabilistic Structured Grammatical Evolution to efficient exploration of the search space. A case study in glucose prediction}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1433--1442}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {genetic algorithms, genetic programming, grammatical evolution, Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726444}, doi = {doi:10.1145/3712256.3726444}, size = {10 pages}, abstract = {People with Type 1 diabetes need to predict their blood glucose levels regularly to keep them within a safe range. Accurate predictions help prevent short-term issues like hypoglycemia and reduce the risk of long-term complications. Evolutionary algorithms have shown potential for this task by generating reliable models for glucose prediction.This work compares four evolutionary approaches: Structured Grammatical Evolution (SGE), a float-based variant (SGEF), and two probabilistic methods, Probabilistic SGE (PSGE) and Co-evolutionary PSGE (Co-PSGE). These methods are tested on their ability to predict glucose levels two hours ahead in individuals with diabetes. Two aspects are examined: predictive performance and the diversity of the phenotypes produced by each approach.Results indicate that SGEF provides statistically better performance than the other methods. Although PSGE and Co-PSGE do not show statistically significant improvements in prediction accuracy, they generate a broader set of solutions and explore more distinct areas of the search space.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(nasra:2025:GECCO, author = {Iyed Nasra and Herve Camus and Ghaith Manita and Amine Dhraief and Ouajdi Korbaa}, title = {Orthogonal Genetic Algorithm for Efficient Delivery Route Planning in {TSP-D}}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1443--1452}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726443}, doi = {doi:10.1145/3712256.3726443}, size = {10 pages}, abstract = {In this study, we propose an advanced Orthogonal Genetic Algorithm (OGA) specifically developed to tackle the Traveling Salesman Problem with Drones (TSP-D), a multifaceted optimization challenge that necessitates precise synchronization between a truck and a drone for effective delivery tasks. The OGA integrates Orthogonal Crossover and Region-Based Mutation strategies, thereby enhancing the algorithm's proficiency in optimizing drone routing in a range of TSP-D scenarios. This novel approach significantly augments the algorithm's adaptability and exploratory capabilities within the intricate search space. Our comprehensive experimental analysis rigorously evaluates the performance of the proposed OGA against established algorithms in a variety of TSP-D instances. The results from these evaluations reveal that our approach substantially surpasses conventional algorithms in terms of both convergence speed and solution quality. This enhanced performance underscores the OGA's efficacy and robustness in optimizing complex paths in TSP-D scenarios.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(nyl\ae{}nder:2025:GECCO, author = {Karoline Nyl\ae{}nder and Aitor Arrieta and Shaukat Ali and Paolo Arcaini}, title = {Search-based Generation of Waypoints for Triggering Self-Adaptations in Maritime Autonomous Vessels}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1453--1461}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726399}, doi = {doi:10.1145/3712256.3726399}, size = {9 pages}, abstract = {Self-adaptation in maritime autonomous vessels (AVs) enables them to adapt their behaviors to address unexpected situations while maintaining dependability requirements. During the design of such AVs, it is crucial to understand and identify the settings that should trigger adaptations, enabling validation of their implementation. To this end, we focus on the navigation software of AVs, which must adapt their behavior during operation through adaptations. AVs often rely on predefined waypoints to guide them along designated routes, ensuring safe navigation. We propose a multi-objective search-based approach, called WPgen, to generate minor modifications to the predefined set of waypoints, keeping them as close as possible to the original waypoints, while causing the AV to navigate inappropriately when navigating with the generated waypoints. WPgen uses NSGA-II as the multi-objective search algorithm with three seeding strategies for its initial population, resulting in three variations of WPgen. We evaluated these variations on three AVs (one overwater tanker and two underwater). We compared the three variations of WPgen with Random Search as the baseline and with each other. Experimental results showed that the effectiveness of these variations varied depending on the AV. Based on the results, we present the research and practical implications of WPgen.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(nuebel:2025:GECCO, author = {Carlo Nuebel and Malte Speidel and Sanaz Mostaghim}, title = {Navigating Path-Influenced Environments using Evolutionary Multi-Objective Optimization}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1462--1470}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {multi-objective optimization, evolutionary algorithms, pathfinding, path-influenced environments, Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726349}, doi = {doi:10.1145/3712256.3726349}, size = {9 pages}, abstract = {This paper explores multi-objective pathfinding in path-influenced environments. These environments contain movable obstacles which can be shifted by the agents. This way, the agents actively change their environment while traversing on their path. Therefore, pathfinding takes on a new dimension. While it has been extensively studied across various domains, finding an optimal path in a path-influenced environment introduces new challenges. In this paper, we propose several real-world inspired problem instances. Then we formally describe this sort of problem as a multi-objective optimization problem and finally evaluate the performance of seven state-of-the-art multi-objective evolutionary algorithms on our problem instances. The results indicate that the evolutionary approach can generate sets of non-dominated solutions for this new problem. The performance of the algorithms in terms of convergence and diversity of the Pareto front highly depends on the way the encountered obstacles are handled, as well as the obstacle distribution on the map. Among the algorithms, AGE-MOEA and SPEA-II demonstrate the best convergence across the majority of problem instances.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(schrum:2025:GECCO, author = {Jacob Schrum and Cody Crosby}, title = {A Quality Diversity Approach to Evolving Model Rockets}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1471--1479}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726338}, doi = {doi:10.1145/3712256.3726338}, size = {9 pages}, abstract = {Model rocketry presents a design task accessible to undergraduates while remaining an interesting challenge. Allowing for variation in fins, nose cones, and body tubes presents a rich design space containing numerous ways to achieve various altitudes. Therefore, when exploring possible designs computationally, it makes sense to apply a method that produces various possibilities for decision-makers to choose from: Quality Diversity (QD). The QD methods MAP-Elites, CMA-ME, and CMA-MAE are applied to model rocket design using the open-source OpenRocket software to characterize the behavior and determine the fitness of evolved designs. Selected rockets were manufactured and launched to evaluate them in the real world. Simulation results demonstrate that CMA-ME produces the widest variety of rocket designs, which is surprising given that CMA-MAE is a more recent method designed to overcome shortcomings with CMA-ME. Real-world testing demonstrates that a wide range of standard and unconventional designs are viable, though issues with the jump from simulation to reality cause some rockets to perform unexpectedly. This paper provides a case study on applying QD to a task accessible to a broader audience than industrial engineering tasks and uncovers unexpected results about the relative performance of different QD algorithms.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(seidelmann:2025:GECCO, author = {Thomas Seidelmann and Sanaz Mostaghim}, title = {Optimization of Unequal-Area Facility Layouts for Mass-Customization Assembly Systems with {AGV} Material Handling}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1480--1488}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726441}, doi = {doi:10.1145/3712256.3726441}, size = {9 pages}, abstract = {Traditional facility layout planning (FLP) typically assumes predictable, static or periodic material flows, which no longer applies to modern mass-customization assembly systems. Optimizing these systems requires a complex integration of unequal-area FLP with the dynamic flexible assembly job-shop scheduling problem (DFA-JSP), and adaptable material handling provided by multiple-load automated guided vehicle (AGV) dispatching. Due to high variability in material flows, stochastic processing times, and dynamic AGV availability, traditional methods fail to address the new problem effectively. This paper introduces the first solution approach to integrate these three NP-hard problems through a combination of multi-objective evolutionary algorithms and advanced dispatching rule systems, thus offering a significant advancement in addressing the planning challenges of mass-customization assembly systems. We evaluate three optimization architectures in 12 configurations and demonstrate that a mutation-only NSGA-II with adaptive parameter adjustment outperforms multi-stage optimization slightly, and cooperative coevolution approaches substantially on this problem. The results suggest that transitioning between early exploration and late exploitation is essential for optimal results, while coevolutionary search space division has little benefit. For the same scenario and budget, the proposed algorithm improves flow time by approximately 5\% and reduces idle time by 9\% compared to the coevolutionary algorithms.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(sa:2025:GECCO, author = {Bruno Sa and Alexandre Oliveira and Miguel Rocha}, title = {Evolutionary Algorithms for Metabolic Transformation through Multi-gene Knockout Optimization}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1489--1496}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {metabolic transformation algorithm, constraint-based metabolic models, evolutionary algorithms, aging, Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726477}, doi = {doi:10.1145/3712256.3726477}, size = {8 pages}, abstract = {The Metabolic Transformation Algorithm (MTA) leverages constraint-based modeling to identify metabolic interventions capable of shifting a biological system from an undesired to a desired state. Its robust extension (rMTA) strengthens predictive accuracy through worst-case scenario analyses and the integration of Minimization Of Metabolic Adjustment (MOMA) algorithm.In this work, we applied Robust Metabolic Transformation Algorithm (rMTA) to an aging-related scenario in Caenorhabditis elegans, focusing on unc-62, a gene implicated in longevity and age-associated metabolic pathways. Building on rMTA derived insights, we developed Evolutionary Algorithms (EAs) that systematically explore combinatorial gene interventions by encoding multigene knockouts as binary vectors and using Robust Transformation Score (rTS) to build the objective function. Through this approach, we uncovered synergistic deletions that significantly outperform single-gene knockouts in redirecting the metabolic network toward a healthier phenotype.By expanding beyond single-gene modifications, our integrated rMTA-EA framework enables a more comprehensive search for metabolic targets that drive phenotype reversion. Although demonstrated here in C. elegans, this method is broadly applicable to other organisms and complex diseases, providing a scalable platform for discovering multi-gene strategies in systems biology and metabolic engineering.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(chien:2025:GECCO, author = {Trinh Van Chien and Bui Trong Duc and Mohammadali Mohammadi and Hien Ngo and Michail Matthaiou}, title = {Differential Evolution for Infeasible Circumstances in Network-Assisted Full-Duplex Cell-Free Massive {MIMO}}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1497--1505}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {full-duplex communication, cell-free massive mimo, differential evolution, spectral efficiency maximization, Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726476}, doi = {doi:10.1145/3712256.3726476}, size = {9 pages}, abstract = {This paper presents an application of differential evolution in optimizing the exploitation of full-duplex communication for Cell-Free Massive Multiple Input Multiple Output (CF-mMIMO), a potential candidate for 6G networks. This paper proposes a new dynamic network-assisted full-duplex CF-mMIMO network, where access points can operate in either half-duplex or full-duplex mode, and each full-duplex access point can serve uplink and downlink users simultaneously. A long-term total spectral efficiency maximization problem is formulated subject to a network operation model and individual spectral efficiency requirements with a limited power budget. Due to the intrinsic nonconvexity and infeasible circumstances where some users might not achieve the rate requirements, we adapt differential evolution to design a low computational complexity algorithm, attaining good power allocation and network operation mode in polynomial time. We further analytically investigate the number of generations required to reach the optimal solution. Numerical results demonstrate the effectiveness of our system design and proposed algorithm over state-of-the-art benchmarks. The network can offer satisfactory service to most users, although several may be unscheduled under harsh conditions.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(zhang:2025:GECCO2, author = {Yu Zhang and Yuehe Zhu and Jiacheng Zhang and Yazhong Luo}, title = {Sequence Optimization of Multispacecraft Multitarget Rendezvous Missions with a Coevolutionary Algorithm}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Roman Kalkreuth and Alexander Brownlee}, pages = {1506--1514}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Real World Applications}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726321}, doi = {doi:10.1145/3712256.3726321}, size = {9 pages}, abstract = {The optimization of multispacecraft multitarget rendezvous sequences is a critical component of mission planning for endeavors such as space debris removal and multiasteroid exploration. This approach has significant implications for improving mission efficiency. a coevolutionary algorithm that synergistically combines a genetic algorithm (GA) with ant colony optimization (ACO) is proposed to solve a multispacecraft multitarget rendezvous sequence planning problem. The proposed algorithm employs a hierarchical strategy for target allocation and sequence planning, seamlessly integrating these two layers into a unified coevolutionary framework. To address the time-dependent challenges in space target rendezvous sequence planning, the ACO algorithm incorporates a time-local search strategy to optimize the rendezvous sequence and rendezvous time. Two local search strategies, which are called tail addition and sequence replanning, are further proposed to address the challenge associated with remaining targets in fuel-constrained missions. The experimental results highlight the effectiveness of the coevolutionary algorithm in solving the multispacecraft multitarget rendezvous sequence planning problem, with performance improvements of 8.4--10.2 percent in optimality over traditional methods. In the modified scenarios of the 12th Global Trajectory Optimization Competition, the algorithm's optimization results are superior to those of the winning team, highlighting its exceptional optimization capabilities.}, notes = {GECCO-2025 RWA A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(ali:2025:GECCO, author = {Aseel Ismael Ali and Edward Keedwell and Ayah Helal}, title = {Learning Grouping Heuristics in Ant Colony Optimization for Combinatorial Problems}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Paola Pellegrini and Ed Keedwell}, pages = {1515--1522}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {ant colony optimization, learning grouping heuristic, combinatorial optimisation problems, learning-based heuristic, bin packing problem, travelling salesman problem, knapsack problem, Swarm Intelligence}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726416}, doi = {doi:10.1145/3712256.3726416}, size = {8 pages}, abstract = {Ant colony optimisation (ACO) has demonstrated good performance on a number of combinatorial optimisation tasks. A recent advance demonstrated the successful addition of a grouping heuristic used information from the objective function to prioritise solutions with full bins. This method increased performance further and established grouping-ACO among the state-of-the-art approaches to bin packing. In this paper, we develop a method to learn and apply decision variable groupings during the ACO algorithm run with no additional information from the objective function. This enables the approach to be generalised to any combinatorial problems for which an ACO representation can be formulated. Experimentation is conducted on a number of instances of the bin packing, knapsack and travelling salesman problems and shows improved performance over standard ACO in all cases, and performance approaching grouping-ACO on the bin packing problem.}, notes = {GECCO-2025 SI A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(asadi:2025:GECCO, author = {Mehrdad Asadi and Ann Nowe and Javad Ghofrani}, title = {Congestion-Aware Multi-Agent Path Planning for Pick-Up and Delivery Tasks}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Paola Pellegrini and Ed Keedwell}, pages = {1523--1531}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Swarm Intelligence}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726339}, doi = {doi:10.1145/3712256.3726339}, size = {9 pages}, abstract = {Mobile robotic systems play a pivotal role in logistics, particularly in warehouse operations, where efficient and collision-free navigation is essential for completing tasks. However, managing a large number of robots often leads to congestion, causing delays and adversely affecting system scalability. This paper proposes a novel online algorithm for solving the Multi-Agent Pickup and Delivery (MAPD) problem. The algorithm addresses local collision detection and global congestion avoidance by integrating a congestion prediction model to enhance process efficiency. A deep neural network is employed to approximate congestion predictions independently of the number of agents, reducing computational complexity. Simulation experiments demonstrate that the proposed approach significantly improves system throughput and scalability, with a notable average doubling of throughput in specific scenarios. The findings provide a foundation for advanced congestion management strategies in multi-agent systems, paving the way for efficient and scalable deployment in logistics and beyond.}, notes = {GECCO-2025 SI A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(brookhouse:2025:GECCO, author = {James Brookhouse and Ayah Helal and Fernando Otero}, title = {An Ensemble Ant Colony Optimization Algorithm with a Hybrid Pheromone Model for Learning Rule Lists}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Paola Pellegrini and Ed Keedwell}, pages = {1532--1539}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Swarm Intelligence}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726427}, doi = {doi:10.1145/3712256.3726427}, size = {8 pages}, abstract = {In this paper, we present an ensemble hybrid pheromone Ant-Miner based algorithm, eAnt-MinerPB+HMA, which benefits from a new hybrid pheromone model to improve the computational and execution time of the algorithm, along with ensemble methods to boost predictive performance. Ant Colony Optimization (ACO) based rule induction algorithms have proven to be successful in producing classification rules. Ensemble methods have also been shown to boost the predictive performance of individual learners, leading to better models. eAnt-MinerPB+HMA creates multiple colonies to build a set of classifiers through feature and instance bagging. eAnt-MinerPB+HMA shows competitive accuracy compared to traditional Ant-Miner variants, while also improving its execution speed-more noticeably in larger data sets.}, notes = {GECCO-2025 SI A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(fujita:2025:GECCO, author = {Shoei Fujita and Ryuki Ishizawa and Hiroyuki Sato and Keiki Takadama}, title = {Adaptive Multi-Population Dynamic Optimization for Multimodal Dynamic Function Optimization}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Paola Pellegrini and Ed Keedwell}, pages = {1540--1548}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {dynamic optimization, swarm intelligence, particle swarm optimization, evolution strategies, hybridization}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726473}, doi = {doi:10.1145/3712256.3726473}, size = {9 pages}, abstract = {To tackle the dynamic optimization problem where the location and number of optimal solutions frequently change, this paper proposes NDSOT (Niching Swarm Dynamic Optimization with TCMA-ES) which can continuously track the multiple moving optimal solutions each of which is generated or eliminated as time goes on. For this purpose, the proposed algorithm integrates Tracking CMA-ES (TCMA-ES) with NMMSO (Niching Migratory Multi-Swarm Optimization), where the former aims to locally track the multiple moving optimal solutions with globally estimating their movement direction, while the latter aims to estimate the number of the multiple optimal solutions by adjusting the number of swarms composed of individuals. The intensive experiments of the three dynamic multimodal functions in 2D and 5D from the Moving Peaks Benchmark (MPB) have revealed that NDSOT succeeded to track the multiple moving and generated/eliminated optimal solutions. In detail, (1) the Offline-Error and Relative Error Distance of NDSOT is lowest and (2) the Peak Found Ratio of NDSOT is highest in comparison with the conventional methods of multiswarm-Quantum Particle Swarm Optimization (QPSO) and TSOPC.}, notes = {GECCO-2025 SI A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(leuzzi:2025:GECCO, author = {Lorenzo Leuzzi and Davide Bacciu and Sabine Hauert and Simon Jones and Andrea Cossu}, title = {Lifelong Evolution of Swarms}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Paola Pellegrini and Ed Keedwell}, pages = {1549--1557}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Swarm Intelligence}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726384}, doi = {doi:10.1145/3712256.3726384}, size = {9 pages}, abstract = {Adapting to task changes without forgetting previous knowledge is a key skill for intelligent systems, and a crucial aspect of lifelong learning. Swarm controllers, however, are typically designed for specific tasks, lacking the ability to retain knowledge across changing tasks. Lifelong learning, on the other hand, focuses on individual agents with limited insights into the emergent abilities of a collective like a swarm. To address this gap, we introduce a lifelong evolutionary framework for swarms, where a population of swarm controllers is evolved in a dynamic environment that incrementally presents novel tasks. This requires evolution to find controllers that quickly adapt to new tasks while retaining knowledge of previous ones, as they may reappear in the future. We discover that the population inherently preserves information about previous tasks, and it can reuse it to foster adaptation and mitigate forgetting. In contrast, the top-performing individual for a given task catastrophically forgets previous tasks. To mitigate this phenomenon, we design a regularization process for the evolutionary algorithm, reducing forgetting in top-performing individuals. Evolving swarms in a lifelong fashion raises fundamental questions on the current state of deep lifelong learning and on the robustness of swarm controllers in dynamic environments.}, notes = {GECCO-2025 SI A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(loi:2025:GECCO, author = {Alessia Loi and Nicolas Bredeche}, title = {Evolving Neural Controllers for Adaptive Visual Pattern Formation by a Swarm of Robots}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Paola Pellegrini and Ed Keedwell}, pages = {1558--1566}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {adaptive pattern formation, swarm robotics, multi-cellular developmental systems, evolutionary robotics, Swarm Intelligence}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726484}, doi = {doi:10.1145/3712256.3726484}, size = {9 pages}, abstract = {In this work, we explore the evolution of neural controllers to coordinate a swarm of robotic agents that dynamically adjust their state to match a target pattern defined at the macroscopic level (e.g. each robot should display a specific color so that an external observer sees a coherent picture from the swarm). Inspired by the multi-cellular flag problem, we compare static and dynamic swarm setups using a sliding puzzle-inspired grid environment. We use evolutionary learning to optimize artificial neural network (ANN) controllers that regulate agent behaviors based on local communication. We analyze the impact of agent density and movement fluidity on pattern formation, and we demonstrate that allowing controlled movement can enhance adaptability while preserving global structure. Post-mortem analyses reveal key differences in learned strategies between static and dynamic configurations regarding generalization to varying swarm density, including validation with real robots. We also reveal how communication speed is critical when the swarm configuration changes over time.}, notes = {GECCO-2025 SI A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(sartorio:2025:GECCO, author = {Vinicius Sartorio and Luigi Feola and Vito Trianni and Jonata Tyska Carvalho}, title = {Minimalist exploration strategies for robot swarms at the edge of chaos}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Paola Pellegrini and Ed Keedwell}, pages = {1567--1576}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Swarm Intelligence}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726311}, doi = {doi:10.1145/3712256.3726311}, size = {10 pages}, abstract = {Effective exploration abilities are fundamental for robot swarms, especially when small, inexpensive robots are employed (e.g., micro- or nano-robots). Random walks are often the only viable choice if robots are too constrained regarding sensors and computation to implement state-of-the-art solutions. However, identifying the best random walk parameterisation may not be trivial. Additionally, variability among robots in terms of motion abilities-a very common condition when precise calibration is not possible-introduces the need for flexible solutions. This study explores how random walks that present chaotic or edge-of-chaos dynamics can be generated. We also evaluate their effectiveness for a simple exploration task performed by a swarm of simulated Kilobots. First, we show how Random Boolean Networks can be used as controllers for the Kilobots, achieving a significant performance improvement compared to the best parameterisation of a L\'{e}vy-modulated Correlated Random Walk. Second, we demonstrate how chaotic dynamics are beneficial to maximise exploration effectiveness. Finally, we demonstrate how the exploration behavior produced by Boolean Networks can be optimized through an Evolutionary Robotics approach while maintaining the chaotic dynamics of the networks achieving 7.6\% of improvement compared to the baseline.}, notes = {GECCO-2025 SI A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(wei:2025:GECCO, author = {Jie Wei and Yuhui Zhang and Wenhong Wei}, title = {{HSEPSO:} A Hierarchical Self-Evolutionary {PSO} Approach for {UAV} Path Planning}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Paola Pellegrini and Ed Keedwell}, pages = {1577--1584}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Swarm Intelligence}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726306}, doi = {doi:10.1145/3712256.3726306}, size = {8 pages}, abstract = {This paper proposes a Hierarchical Self-Evolutionary PSO (HSEPSO) Approach for UAV Path Planning to address the challenges faced by traditional Particle Swarm Optimization (PSO), such as high sensitivity to parameters, the tendency to become trapped in local optima, and slow convergence in later stages. Additionally, existing improvements to PSO lack the ability to dynamically adjust evolution strategies based on the current state of particles. HSEPSO employs a hybrid clustering strategy combining K-Means and DB-SCAN for population initialization, followed by population division based on clustering results. This ensures diversity within the population while enabling particles to focus their search on regions more likely to contain the optimal solution. The algorithm also dynamically adjusts the learning factors and inertia weights through a nonlinear adaptive update strategy, effectively balancing global search and local exploitation. Moreover, based on the real-time state of the particles, HSEPSO incorporates different evolutionary strategies to accelerate convergence, optimize the search for solutions, and enhance algorithm robustness. Experimental results demonstrate that, compared to traditional PSO and other improved algorithms (such as MFIPSO, SDPSO, and SA2PSO), HSEPSO shows notable improvements in optimization performance, convergence speed, and robustness.}, notes = {GECCO-2025 SI A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(adak:2025:GECCO, author = {Sumit Adak and Carsten Witt}, title = {Improved Runtime Analysis of a Multi-Valued Compact Genetic Algorithm on Two Generalized {OneMax} Problems}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Christine Zarges and Dirk Sudholt}, pages = {1585--1593}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {estimation of distribution algorithms, multi-valued compact genetic algorithm, genetic drift, OneMax, Theory}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726353}, doi = {doi:10.1145/3712256.3726353}, size = {9 pages}, abstract = {Recent research in the runtime analysis of estimation of distribution algorithms (EDAs) has focused on univariate EDAs for multi-valued decision variables. In particular, the runtime of the multi-valued cGA (r-cGA) and UMDA on multi-valued functions has been a significant area of study. Adak and Witt (PPSN 2024) and Hamano et al. (ECJ 2024) independently performed a first runtime analysis of the r-cGA on the r-valued OneMax function (r-OneMax). Adak and Witt also introduced a different r-valued OneMax function called G-OneMax. However, for that function, only empirical results were provided so far due to the increased complexity of its runtime analysis, since r-OneMax involves categorical values of two types only, while G-OneMax encompasses all possible values.In this paper, we present the first theoretical runtime analysis of the r-cGA on the G-OneMax function. We demonstrate that the runtime is O(nr3 log2 n log r) with high probability. Additionally, we refine the previously established runtime analysis of the r-cGA on r-OneMax, improving the previous bound to O(nr log n log r), which improves the state of the art by an asymptotic factor of log n and is tight for the binary case. Moreover, we for the first time include the case of frequency borders.}, notes = {GECCO-2025 THEORY A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(benford:2025:GECCO, author = {Alistair Benford and Per Kristian Lehre}, title = {A General Upper Bound for the Runtime of a Coevolutionary Algorithm on Impartial Combinatorial Games}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Christine Zarges and Dirk Sudholt}, pages = {1594--1603}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {Theory}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726437}, doi = {doi:10.1145/3712256.3726437}, size = {10 pages}, abstract = {Due to their complex dynamics, combinatorial games are a key test case and application for algorithms that train game playing agents. Among those algorithms that train using self-play are coevolutionary algorithms (CoEAs). However, the successful application of CoEAs for game playing is difficult due to pathological behaviours such as cycling, an issue especially critical for games with intransitive payoff landscapes.Insight into how to design CoEAs to avoid such behaviours can be provided by runtime analysis. In this paper, we push the scope of runtime analysis for CoEAs to combinatorial games, proving a general upper bound for the number of simulated games needed for UMDA to discover (with high probability) an optimal strategy. This result applies to any impartial combinatorial game, and for many games the implied bound is polynomial or quasipolynomial as a function of the number of game positions. After proving the main result, we provide several applications to simple well-known games: Nim, Chomp, Silver Dollar, and Turning Turtles. As the first runtime analysis for CoEAs on combinatorial games, this result is a critical step towards a comprehensive theoretical framework for coevolution.}, notes = {GECCO-2025 THEORY A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(dang:2025:GECCO, author = {Duc-Cuong Dang and Andre Opris and Dirk Sudholt}, title = {Why Dominance Is Not Enough: Lessons from Practical Evolutionary Multi-Objective Algorithms}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Christine Zarges and Dirk Sudholt}, pages = {1604--1612}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {runtime analysis, evolutionary multiobjective optimisation, Theory}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726414}, doi = {doi:10.1145/3712256.3726414}, size = {9 pages}, abstract = {Practical EMO algorithms like NSGA-II, NSGA-III, and SMS-EMOA combine the dominance relation with diversity criteria to identify promising solutions. Despite many success stories, their theoretical foundation remains underdeveloped, with key questions still unanswered-such as which information obtained throughout the evolution is critical for their success.We explore the limitations of the information provided by the dominance relation between search points encountered so far. We construct an artificial problem with a small Pareto set where almost all pairs of search points are incomparable. For this problem, we prove that any black-box EMO algorithm that only relies on the dominance relation for making decisions and only use variation operators that are invariant to bit values, fails spectacularly, requiring exponential time with high probability. In stark contrast, NSGA-II, NSGA-III, and SMS-EMOA efficiently cover the Pareto front in expected quadratic time by incorporating additional information, such as objective values. Our results highlight the superiority of practical EMO algorithms and the necessity of using information beyond dominance for effective multi-objective optimisation.}, notes = {GECCO-2025 THEORY A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(lengler:2025:GECCO, author = {Johannes Lengler and Aneta Neumann and Frank Neumann}, title = {Runtime Analysis of Evolutionary Multitasking for Classical Benchmark Problems}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Christine Zarges and Dirk Sudholt}, pages = {1613--1621}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {evolutionary multitasking, runtime analysis, theory}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726369}, doi = {doi:10.1145/3712256.3726369}, size = {9 pages}, abstract = {Evolutionary multitasking has gained significant attention in the evolutionary computation literature in recent years. Here an evolutionary algorithm is used to compute good or optimal solutions for not just a single but several (possibly related) tasks. We provide a first runtime analysis of evolutionary multitask algorithms and investigate generalized versions of OneMax, LeadingOnes, and Jump which are classical benchmark functions frequently studied in the area of runtime analysis. Our theoretical investigations point out significant speed ups when using evolutionary multitasking instead of several runs of the classical (1+1) EA. In particular, our analysis reveals how progress is shared between the different tasks using uniform crossover in an evolutionary multitasking algorithm. We complement our asymptotic theoretical analysis by experimental investigations which provide further insights into the actual speed ups dependent on the similarity of the given tasks for realistic problem sizes.}, notes = {GECCO-2025 THEORY A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(ma:2025:GECCO3, author = {Yuxuan Ma and Pietro S. Oliveto and John Warwicker}, title = {Random Gradient Hyper-heuristics Can Learn to Escape Local Optima in Multimodal Optimisation}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Christine Zarges and Dirk Sudholt}, pages = {1622--1630}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {hyper-heuristics, parameter adaptation, multimodal optimisation, runtime analysis, theory}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726406}, doi = {doi:10.1145/3712256.3726406}, size = {9 pages}, abstract = {Selection hyper-heuristics (SHHs) select from a set of low-level heuristics which to apply during the optimisation process. One such approach, namely the random gradient SHH, which continues to apply a randomly selected heuristic as long as it remains successful, has been shown to be able to effectively select heuristics leading to optimal expected runtimes on a range of unimodal functions. In this work, we extend the analysis of the random gradient SHH to multimodal optimisation problems to assess their performance at escaping from local optima. We consider the TwoRates benchmark function which includes several consecutive local optima separated by gaps of two alternating different sizes. The function was recently introduced to assess the performance of the flex-EA that uses an archive to store and re-apply the two most suitable Randomized Local Search (RLSk) operators to make the jumps of different lengths. We show that the SHH can optimise the function considerably faster by identifying and consecutively re-applying the single best heuristic to overcome all of the local optima. This performance also holds when the set of low-level heuristics contains all the n possible RLSk operators, where n is the problem size.}, notes = {GECCO-2025 THEORY A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, ) @inproceedings(opris:2025:GECCO, author = {Andre Opris and Sebastian Sonntag and Dirk Sudholt}, title = {A Royal Road Function for Permutation Spaces: an Example Where Order Crossover is Provably Essential}, booktitle = {Proceedings of the 2025 Genetic and Evolutionary Computation Conference}, year = {2025}, editor = {Christine Zarges and Dirk Sudholt}, pages = {1631--1640}, address = {Malaga, Spain}, series = {GECCO '25}, month = {14-18 July}, organisation = {SIGEVO}, publisher = {Association for Computing Machinery}, publisher_address = {New York, NY, USA}, note = {}, keywords = {crossover, recombination, permutation spaces, runtime analysis, Theory}, ISBN13 = {979-8-4007-1465-8}, url = {https://doi.org/10.1145/3712256.3726403}, doi = {doi:10.1145/3712256.3726403}, size = {10 pages}, abstract = {Permutation spaces represent a wide range of important problems in domains such as scheduling, routing, sequencing, and assignment. Despite the frequent application of evolutionary algorithms to permutation-based problems, the theory of evolutionary computing in permutation spaces is in its infancy. Many fundamental questions remain open, particularly regarding the effectiveness of various mutation and crossover operators designed for permutation spaces. While there is a substantial body of runtime analyses demonstrating the benefits of crossover in pseudo-Boolean optimisation, there is no such work for permutation spaces.We present the first example of a permutation problem in which the use of crossover is proven to be beneficial. Mutation-only evolutionary algorithms, such as the (1+1) EA with swaps, exchanges, or jumps as mutation operators, require exponential time to find the global optimum with high probability. In contrast, an island model and a (mu+1) EA with fitness sharing both leverage order crossover to achieve polynomial runtimes. These stark performance differences highlight the potential advantages of crossover and pave the way for systematically exploring its effectiveness in permutation spaces.}, notes = {GECCO-2025 THEORY A Recombination of the 34th International Conference on Genetic Algorithms (ICGA) and the 30th Annual Genetic Programming Conference (GP)}, )