Yifan He
ABSTRACT
Fitness landscape analysis (FLA) refers to a set of techniques to characterise optimisation problems. This paper presents an FLA of three types of genetic programming (GP) benchmarks: parity, symbolic regression, and artificial ant. We applied a modern graph-based FLA tool called Local Optima Networks and several classical FLA metrics (fitness distance correlation, neutrality, and ruggedness measures) to study the tree-based GP search spaces. Our analysis shows that the search spaces for all problems contain many local optima and are highly deceptive. The parity problems are highly rugged and neutral. Conversely, the problems of symbolic regression are less rugged and neutral. Finally, the artificial ant problem is highly rugged but less neutral. Our results indicate that a mutation in deep nodes makes finding the global optimum difficult.
- Jason Adair, Gabriela Ochoa, and Katherine M Malan. 2019. Local optima networks for continuous fitness landscapes. In Proceedings of the Genetic and Evolutionary Computation Conference Companion. 1407--1414.Google Scholar
Digital Library
- Anna Bosman. 2019. Fitness Landscape Analysis of Feed-Forward Neural Networks. Ph.D. Dissertation. University of Pretoria.Google Scholar
- Rebeka Čorić, Mateja Ðumić, and Domagoj Jakobović. 2021. Genetic programming hyperheuristic parameter configuration using fitness landscape analysis. Applied Intelligence 51 (2021), 7402--7426.Google ScholarDigital Library
- Marko Durasevic, Domagoj Jakobovic, Marcella Scoczynski Ribeiro Martins, Stjepan Picek, and Markus Wagner. 2020. Fitness landscape analysis of dimensionally-aware genetic programming featuring feynman equations. In Parallel Problem Solving from Nature-PPSN XVI: 16th International Conference, PPSN 2020, Leiden, The Netherlands, September 5--9, 2020, Proceedings, Part II 16. Springer, 111--124.Google Scholar
- Thomas Elsken, Jan Hendrik Metzen, and Frank Hutter. 2019. Neural architecture search: A survey. The Journal of Machine Learning Research 20, 1 (2019), 1997--2017.Google ScholarDigital Library
- Qinglan Fan, Ying Bi, Bing Xue, and Mengjie Zhang. 2022. Genetic programming for feature extraction and construction in image classification. Applied Soft Computing 118 (2022), 108509.Google ScholarDigital Library
- Terry Jones, Stephanie Forrest, et al. 1995. Fitness distance correlation as a measure of problem difficulty for genetic algorithms.. In ICGA, Vol. 95. 184--192.Google Scholar
- John R Koza. 1990. Genetically breeding populations of computer programs to solve problems in artificial intelligence. In [1990] Proceedings of the 2nd International IEEE Conference on Tools for Artificial Intelligence. IEEE, 819--827.Google Scholar
- John R Koza. 1994. Genetic programming as a means for programming computers by natural selection. Statistics and computing 4 (1994), 87--112.Google Scholar
- Joel Lehman, Kenneth O Stanley, et al. 2008. Exploiting open-endedness to solve problems through the search for novelty.. In ALIFE. 329--336.Google Scholar
- Alexander Loginov, Malcolm Heywood, and Garnett Wilson. 2021. Stock selection heuristics for performing frequent intraday trading with genetic programming. Genetic Programming and Evolvable Machines 22 (2021), 35--72.Google ScholarDigital Library
- Hui Lu, Rongrong Zhou, Zongming Fei, and Chongchong Guan. 2019. Spatial-domain fitness landscape analysis for combinatorial optimization. Information Sciences 472 (2019), 126--144.Google ScholarCross Ref
- Katherine Mary Malan. 2021. A survey of advances in landscape analysis for optimisation. Algorithms 14, 2 (2021), 40.Google ScholarCross Ref
- Abdul Manazir and Khalid Raza. 2019. Recent developments in cartesian genetic programming and its variants. ACM Computing Surveys (CSUR) 51, 6 (2019), 1--29.Google ScholarDigital Library
- Jean-Baptiste Mouret and Jeff Clune. 2015. Illuminating search spaces by mapping elites. arXiv preprint arXiv:1504.04909 (2015).Google Scholar
- Quang Uy Nguyen, Cong Doan Truong, Xuan Hoai Nguyen, and Michael O'Neill. 2013. Guiding function set selection in genetic programming based on fitness landscape analysis. In Proceedings of the 15th annual conference companion on Genetic and evolutionary computation. 149--150.Google ScholarDigital Library
- Gabriela Ochoa and Nadarajen Veerapen. 2022. Neural Architecture Search: A Visual Analysis. In Parallel Problem Solving from Nature-PPSN XVII: 17th International Conference, PPSN 2022, Dortmund, Germany, September 10--14, 2022, Proceedings, Part I. Springer, 603--615.Google Scholar
- Gabriela Ochoa, Sébastien Verel, Fabio Daolio, and Marco Tomassini. 2014. Local optima networks: A new model of combinatorial fitness landscapes. Recent advances in the theory and application of fitness landscapes (2014), 233--262.Google Scholar
- Nuno M Rodrigues, Katherine M Malan, Gabriela Ochoa, Leonardo Vanneschi, and Sara Silva. 2022. Fitness landscape analysis of convolutional neural network architectures for image classification. Information Sciences 609 (2022), 711--726.Google ScholarDigital Library
- Karel Slanỳ and Lukáš Sekanina. 2007. Fitness landscape analysis and image filter evolution using functional-level cgp. In Genetic Programming: 10th European Conference, EuroGP 2007, Valencia, Spain, April 11--13, 2007. Proceedings 10. Springer, 311--320.Google ScholarCross Ref
- Marco Tomassini, Leonardo Vanneschi, Philippe Collard, and Manuel Clergue. 2005. A study of fitness distance correlation as a difficulty measure in genetic programming. Evolutionary computation 13, 2 (2005), 213--239.Google Scholar
- Nguyen Quang Uy, Nguyen Xuan Hoai, Michael O'Neill, Robert I McKay, and Edgar Galván-López. 2011. Semantically-based crossover in genetic programming: application to real-valued symbolic regression. Genetic Programming and Evolvable Machines 12 (2011), 91--119.Google ScholarDigital Library
- Leonardo Vanneschi, Yuri Pirola, Giancarlo Mauri, Marco Tomassini, Philippe Collard, and Sébastien Verel. 2012. A study of the neutrality of Boolean function landscapes in genetic programming. Theoretical Computer Science 425 (2012), 34--57.Google ScholarDigital Library
- Leonardo Vanneschi, Marco Tomassini, Philippe Collard, and Sébastien Vérel. 2006. Negative slope coefficient: A measure to characterize genetic programming fitness landscapes. In Genetic Programming: 9th European Conference, EuroGP 2006, Budapest, Hungary, April 10--12, 2006. Proceedings 9. Springer, 178--189.Google ScholarDigital Library
- Leonardo Vanneschi, Marco Tomassini, Philippe Collard, Sébastien Vérel, Yuri Pirola, and Giancarlo Mauri. 2007. A comprehensive view of fitness landscapes with neutrality and fitness clouds. In Genetic Programming: 10th European Conference, EuroGP 2007, Valencia, Spain, April 11--13, 2007. Proceedings 10. Springer, 241--250.Google ScholarCross Ref
- Nadarajen Veerapen, Fabio Daolio, and Gabriela Ochoa. 2017. Modelling genetic improvement landscapes with local optima networks. In Proceedings of the Genetic and Evolutionary Computation Conference Companion. 1543--1548.Google ScholarDigital Library
- Nadarajen Veerapen and Gabriela Ochoa. 2018. Visualising the global structure of search landscapes: genetic improvement as a case study. Genetic programming and evolvable machines 19, 3 (2018), 317--349.Google Scholar
- Sébastien Verel, Fabio Daolio, Gabriela Ochoa, and Marco Tomassini. 2012. Local optima networks with escape edges. In Artificial Evolution: 10th International Conference, Evolution Artificielle, EA 2011, Angers, France, October 24--26, 2011, Revised Selected Papers 10. Springer, 49--60.Google Scholar
- Kaizhong Zhang and Dennis Shasha. 1989. Simple fast algorithms for the editing distance between trees and related problems. SIAM journal on computing 18, 6 (1989), 1245--1262.Google Scholar
- Yuyang Zhou and Ferrante Neri. 2022. A Fitness Landscape Analysis of the LeNet-5 Loss Function. In 2022 IEEE Symposium Series on Computational Intelligence (SSCI). IEEE, 352--359.Google ScholarCross Ref
Index Terms
- Fitness Landscape Analysis of Genetic Programming Search Spaces with Local Optima Networks
Recommendations
Communities of Local Optima as Funnels in Fitness Landscapes
GECCO '16: Proceedings of the Genetic and Evolutionary Computation Conference 2016We conduct an analysis of local optima networks extracted from fitness landscapes of the Kauffman NK model under iterated local search. Applying the Markov Cluster Algorithm for community detection to the local optima networks, we find that the ...
Bag of local landscape features for fitness landscape analysis
Evolutionary computation is a research field dealing with black-box and complex optimization problems whose fitness landscapes are usually unknown in advance. It is difficult to select an appropriate evolutionary algorithm and parameters for a given ...
Towards population-based fitness landscape analysis using local optima networks
GECCO '21: Proceedings of the Genetic and Evolutionary Computation Conference CompanionA fitness landscape describes the interaction of a search domain, a cost function on designs drawn from the domain, and a neighbourhood function defining the adjacency of designs --- induced by the optimisation method used. Fitness landscapes can be ...
Comments