ABSTRACT
Estimation of distribution genetic programming (EDA-GP) algorithms are metaheuristics where sampling new solutions from a learned probabilistic model replaces the standard mutation and recombination operators of genetic programming (GP). This paper presents DAE-GP, a new EDA-GP which uses denoising autoencoder long short-term memory networks (DAE-LSTMs) as probabilistic model. DAE-LSTMs are artificial neural networks that first learn the properties of a parent population by mapping promising candidate solutions to a latent space and reconstructing the candidate solutions from the latent space. The trained model is then used to sample new offspring solutions. We show on a generalization of the royal tree problem that DAE-GP outperforms standard GP and that performance differences increase with higher problem complexity. Furthermore, DAE-GP is able to create offspring with higher fitness from a learned model in comparison to standard GP. We believe that the key reason for the high performance of DAE-GP is that we do not impose any assumptions about the relationships between learned variables which is different to previous EDA-GP models. Instead, DAE-GP flexibly identifies and models relevant dependencies of promising candidate solutions.
- Yoshua Bengio, Li Yao, Guillaume Alain, and Pascal Vincent. 2013. Generalized Denoising Auto-Encoders as Generative Models. Advances on Neural Information Processing Systems (NIPS'13) 26 (2013), 899--907.Google Scholar
- Alexander W. Churchill, Siddharth Sigtia, and Chrisantha Fernando. 2014. A denoising autoencoder that guides stochastic search. Technical Report. arXiv:1404.1614Google Scholar
- Matej Črepinšek, Shih-Hsi Liu, and Marjan Mernik. 2013. Exploration and Exploitation in Evolutionary Algorithms: A Survey. ACM Comput. Surv. 45, 3 (2013), 33. Google ScholarDigital Library
- Jason M Daida, Hsiaolei Li, Ricky Tang, and Adam M Hilss. 2003. What Makes a Problem GP-Hard? Validating a Hypothesis of Structural Causes. In Proceedings of the Genetic and Evolutionary Computation Conference (GECCO'03). Springer, Berlin, Heidelberg, 1665--1677.Google ScholarCross Ref
- Félix Antoine Fortin, François Michel De Rainville, Marc André Gardner, Marc Parizeau, and Christian Gagńe. 2012. DEAP: Evolutionary algorithms made easy. Journal of Machine Learning Research 13, 1 (2012), 2171--2175.Google ScholarDigital Library
- Felix A Gers, Jürgen Schmidhuber, and Fred Cummins. 2000. Learning to forget: Continual prediction with LSTM. Neural Computation 12, 10 (2000), 2451--2471.Google ScholarDigital Library
- Yoshihiko Hasegawa and Hitoshi Iba. 2006. Estimation of Bayesian network for program generation. In Proceedings of the Third Asian-Pacific workshop on Genetic Programming. Hanoi, Vietnam, 35--46.Google Scholar
- Yoshihiko Hasegawa and Hitoshi Iba. 2008. A Bayesian Network Approach to Program Generation. IEEE Transactions on Evolutionary Computation 12, 6 (2008), 750--764. Google ScholarDigital Library
- Erik Hemberg, Kalyan Veeramachaneni, James McDermott, Constantin Berzan, and Una-May O'Reilly. 2012. An investigation of local patterns for estimation of distribution genetic programming. In Proceedings of the Genetic and Evolutionary Computation Conference (GECCO '12). ACM, Philadelphia, USA, 767--774. Google ScholarDigital Library
- Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long Short-Term Memory. Neural Computation 9, 8 (1997), 1735--1780. Google ScholarDigital Library
- John H. Holland. 1975. Adaptation in natural and artificial systems: An introductory analysis with applications to biology, control, and artificial intelligence. University of Michigan Press, Ann Arbor, Michigan, USA.Google Scholar
- Terry Jones and Stephanie Forrest. 1995. Fitness Distance Correlation as a Measure of Problem Difficulty for Genetic Algorithms. In Proceedings of the 6th International Conference on Genetic Algorithms. Morgan Kaufmann Publishers, San Francisco, CA, USA, 184--192. Google ScholarDigital Library
- Kangil Kim, Yin Shan, Xuan Hoai Nguyen, and R. I. McKay. 2014. Probabilistic model building in genetic programming: a critical review. Genetic Programming and Evolvable Machines 15, 2 (2014), 115--167. Google ScholarDigital Library
- Diederik P Kingma and Jimmy Ba. 2015. Adam: A method for stochastic optimization. In International Conference on Learning Representations (ICLR'15). San Diego, CA, USA.Google Scholar
- John R. Koza. 1992. Genetic Programming: On the programming of computers by means of natural selection. MIT press, Cambridge, London.Google ScholarDigital Library
- Joseph B. Kruskal. 1983. An Overview of Sequence Comparison: Time Warps, String Edits, and Macromolecules. Society of Industrial and Applied Mathematics (SIAM) Review 25, 2 (1983), 201--237. Google ScholarDigital Library
- Sean Luke. 2000. Two fast tree-creation algorithms for genetic programming. IEEE Transactions on Evolutionary Computation 4, 3 (2000), 274--283. Google ScholarDigital Library
- Graham Neubig, Chris Dyer, Yoav Goldberg, Austin Matthews, Waleed Ammar, Antonios Anastasopoulos, Miguel Ballesteros, David Chiang, Daniel Clothiaux, Trevor Cohn, Kevin Duh, Manaal Faruqui, Cynthia Gan, Dan Garrette, Yangfeng Ji, Lingpeng Kong, Adhiguna Kuncoro, Gaurav Kumar, Chaitanya Malaviya, Paul Michel, Yusuke Oda, Matthew Richardson, Naomi Saphra, Swabha Swayamdipta, and Pengcheng Yin. 2017. DyNet: The Dynamic Neural Network Toolkit. Technical Report. arXiv:CoRR abs/1701.03980Google Scholar
- Christopher Olah. 2015. Understanding LSTM Networks. (2015). http://colah.github.io/posts/2015-08-Understanding-LSTMs/Google Scholar
- Martin Pelikan, Mark W Hauschild, and Fernando G Lobo. 2012. Introduction to estimation of distribution algorithms. Missouri Estimation of Distribution Algorithms Laboratory (MEDAL), Report Nr. 2012003 (2012).Google Scholar
- Riccardo Poli and Nicholas Freitag McPhee. 2008. A Linear Estimation-of-Distribution GP system. In Proceedings of the 11th European Conference on Genetic Programming (EuroGP'08). Springer, Neapel, Italy, 206--217.Google ScholarDigital Library
- Malte Probst. 2015. Denoising Autoencoders for fast Combinatorial Black Box Optimization. Technical Report. (2015). arXiv:1503.01954Google Scholar
- Bill Punch, Doug Zongker, and Erik Goodman. 1996. The Royal Tree Problem, a Benchmark for Single and Multi-population Genetic Programming. In Advances in Genetic Programming II, Peter J. Angeline and Kenneth E. Kinnear Jr. (Eds.). MIT Press, Cambridge, MA, USA, 299--316.Google Scholar
- A Ratle and M Sebag. 2001. Avoiding the bloat with probabilistic grammar-based genetic programming. In 5th International Conference on Artificial Evolution (EA'01). Springer, Le Creusot, France, 255--266.Google Scholar
- Franz Rothlauf. 2011. Design of Modern Heuristics: Principles and Application (1st ed.). Springer, Berlin, Heidelberg. Google ScholarCross Ref
- Rafal Salustowicz and Jürgen Schmidhuber. 1997. Probabilistic incremental program evolution. Evolutionary Computation 5, 2 (1997), 123--141. Google ScholarDigital Library
- Yin Shan, Robert McKay, Daryl Essam, and Hussein Abbass. 2006. A survey of probabilistic model building genetic programming. In Scalable Optimization via Probabilistic Modeling, M Pelikan, K Sastry, and E CantúPaz (Eds.). Springer, Berlin, Heidelberg, 121--160.Google Scholar
- Dominik Sobania and Franz Rothlauf. 2020. Challenges of Program Synthesis with Grammatical Evolution. In Proceedings of the 23rd European Conference on Genetic Programming (EuroGP'20). Springer, Sevilla, Spain (forthcoming).Google ScholarCross Ref
- Léo Françoso Dal Piccol Sotto and Vinícius Veloso de Melo. 2017. A Probabilistic Linear Genetic Programming with Stochastic Context-Free Grammar for solving Symbolic Regression problems. In Proceedings of the Genetic and Evolutionary Computation Conference (GECCO'17). ACM, Berlin, Germany, 1017--1024.Google ScholarDigital Library
- Nitish Srivastava, Elman Mansimov, and Ruslan Salakhutdinov. 2015. Unsupervised learning of video representations using LSTMs. In Proceedings of the 32nd International Conference on Machine Learning (ICML'15). ACM, Lille, France, 843--852. Google ScholarDigital Library
- D. Thierens. 1999. Scalability problems of simple genetic algorithms. Evolutionary computation 7, 4 (1999), 331--352. Google ScholarDigital Library
- Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention Is All You Need. Advances in Neural Information Processing Systems (NIPS) 30 (2017), 5998--6008.Google Scholar
- Pascal Vincent, Hugo Larochelle, Yoshua Bengio, and Pierre Antoine Manzagol. 2008. Extracting and composing robust features with denoising autoencoders. In Proceedings of the 25th International Conference on Machine Learning (ICML'08). ACM, Helsinki, Finland, 1096--1103. Google ScholarDigital Library
- Paul J. Werbos. 1988. Generalization of backpropagation with application to a recurrent gas market model. Neural Networks 1, 4 (1988), 339--356. Google ScholarCross Ref
- Pak-Kan Wong, Leung-Yau Lo, Man-Leung Wong, and Kwong-Sak Leung. 2014. Grammar-Based Genetic Programming with Bayesian Network. In IEEE Congress on Evolutionary Computation (CEC'14). IEEE, Beijing, China, 739--746.Google Scholar
- Pak-Kan Wong, Leung-Yau Lo, Man-Leung Wong, and Kwong-Sak Leung. 2014. Grammar-Based Genetic Programming with Dependence Learning and Bayesian Network Classifier. In Proceedings of the Genetic and Evolutionary Computation Conference (GECCO'14). ACM, Vancouver, Canada, 959--966. Google ScholarDigital Library
- K Yanai and H Iba. 2003. Estimation of distribution programming based on Bayesian network. In IEEE Congress on Evolutionary Computation (CEC '03). IEEE, Canberra, Australia, 1618--1625. Google ScholarCross Ref
Index Terms
- DAE-GP: denoising autoencoder LSTM networks as probabilistic models in estimation of distribution genetic programming
Recommendations
Denoising autoencoder genetic programming for real-world symbolic regression
GECCO '22: Proceedings of the Genetic and Evolutionary Computation Conference CompanionDenoising Autoencoder Genetic Programming (DAE-GP) is a novel neural-network based estimation of distribution genetic programming algorithm that uses denoising autoencoder long short-term memory networks as probabilistic model to replace the standard ...
Small Solutions for Real-World Symbolic Regression Using Denoising Autoencoder Genetic Programming
Genetic ProgrammingAbstractDenoising Autoencoder Genetic Programming (DAE-GP) is a model-based evolutionary algorithm that uses denoising autoencoder long short-term memory networks as probabilistic model to replace the standard recombination and mutation operators of ...
HMXT-GP: an information-theoretic approach to genetic programming that maintains diversity
SAC '11: Proceedings of the 2011 ACM Symposium on Applied ComputingThis paper applies a recent information--theoretic approach to controlling Genetic Algorithms (GAs) called HMXT to tree--based Genetic Programming (GP). HMXT, in a GA domain, requires the setting of selection thresholds in a population and the ...
Comments