ABSTRACT
When using genetic programming for program synthesis, we are usually constrained by a computational budget measured in program executions during evolution. The computational budget is influenced by the choice of population size and number of generations per run leading to a trade-off between both possibilities. To better understand this trade-off, we analyze the effects of different combinations of population sizes and number of generations on performance. Further, we analyze how the use of different variation operators affects this trade-off. We conduct experiments on a range of common program synthesis benchmarks and find that using larger population sizes lead to a better search performance. Additionally, we find that using high probabilities for crossover and mutation lead to higher success rates. Focusing on only crossover or using only mutation usually leads to lower search performance. In summary, we find that large populations combined with high mutation and crossover rates yield highest GP performance for program synthesis approaches.
- Ryan Boldi, Martin Briesch, Dominik Sobania, Alexander Lalejini, Thomas Helmuth, Franz Rothlauf, Charles Ofria, and Lee Spector. 2023. Informed Down-Sampled Lexicase Selection: Identifying productive training cases for efficient problem solving. arXiv preprint arXiv:2301.01488 (2023).Google Scholar
- David Fagan, Michael Fenton, and Michael O'Neill. 2016. Exploring position independent initialisation in grammatical evolution. In 2016 IEEE Congress on Evolutionary Computation (CEC). IEEE, 5060--5067.Google ScholarDigital Library
- Michael Fenton, James McDermott, David Fagan, Stefan Forstenlechner, Erik Hemberg, and Michael O'Neill. 2017. Ponyge2: Grammatical evolution in python. In Proceedings of the Genetic and Evolutionary Computation Conference Companion. 1194--1201.Google ScholarDigital Library
- Austin J Ferguson, Jose Guadalupe Hernandez, Daniel Junghans, Alexander Lalejini, Emily Dolson, and Charles Ofria. 2020. Characterizing the effects of random subsampling on lexicase selection. Genetic Programming Theory and Practice XVII (2020), 1--23.Google Scholar
- Stefan Forstenlechner, David Fagan, Miguel Nicolau, and Michael O'Neill. 2017. A grammar design pattern for arbitrary program synthesis problems in genetic programming. In Genetic Programming: 20th European Conference, EuroGP 2017, Amsterdam, The Netherlands, April 19--21, 2017, Proceedings 20. Springer, 262--277.Google ScholarCross Ref
- Stefan Forstenlechner, Miguel Nicolau, David Fagan, and Michael O'Neill. 2016. Grammar design for derivation tree based genetic programming systems. In Genetic Programming: 19th European Conference, EuroGP 2016, Porto, Portugal, March 30-April 1, 2016, Proceedings 19. Springer, 199--214.Google ScholarCross Ref
- Thomas Helmuth and Peter Kelly. 2021. PSB2: the second program synthesis benchmark suite. In Proceedings of the Genetic and Evolutionary Computation Conference. 785--794.Google ScholarDigital Library
- Thomas Helmuth, Nicholas Freitag McPhee, and Lee Spector. 2018. Program synthesis using uniform mutation by addition and deletion. In Proceedings of the Genetic and Evolutionary Computation Conference. 1127--1134.Google ScholarDigital Library
- Thomas Helmuth and Lee Spector. 2015. General program synthesis benchmark suite. In Proceedings of the 2015 Annual Conference on Genetic and Evolutionary Computation. 1039--1046.Google ScholarDigital Library
- Thomas Helmuth and Lee Spector. 2022. Problem-Solving Benefits of Down-Sampled Lexicase Selection. Artificial Life 27, 3--4 (03 2022), 183--203.Google ScholarCross Ref
- Jose Guadalupe Hernandez, Alexander Lalejini, Emily Dolson, and Charles Ofria. 2019. Random subsampling improves performance in lexicase selection. In Proceedings of the Genetic and Evolutionary Computation Conference Companion. 2028--2031.Google ScholarDigital Library
- R John. 1992. Koza. Genetic Programming: On the Programming of Computers by Means of Natural Selection.Google Scholar
- Alexander Lalejini and Charles Ofria. 2019. Tag-accessed memory for genetic programming. In Proceedings of the Genetic and Evolutionary Computation Conference Companion. 346--347.Google ScholarDigital Library
- Sean Luke and Lee Spector. 1997. A comparison of crossover and mutation in genetic programming. Genetic Programming 97 (1997), 240--248.Google Scholar
- Sean Luke and Lee Spector. 1998. A revised comparison of crossover and mutation in genetic programming. Genetic Programming 98 (1998), 208--213.Google Scholar
- Conor Ryan, John James Collins, and Michael O Neill. 1998. Grammatical evolution: Evolving programs for an arbitrary language. In Genetic Programming: First European Workshop, EuroGP'98 Paris, France, April 14--15, 1998 Proceedings 1. Springer, 83--96.Google ScholarCross Ref
- Dominik Sobania, Martin Briesch, and Franz Rothlauf. 2022. Choose your programming copilot: a comparison of the program synthesis performance of github copilot and genetic programming. In Proceedings of the Genetic and Evolutionary Computation Conference. 1019--1027.Google ScholarDigital Library
- Dominik Sobania, Dirk Schweim, and Franz Rothlauf. 2022. A comprehensive survey on program synthesis with evolutionary algorithms. IEEE Transactions on Evolutionary Computation (2022).Google ScholarCross Ref
- Lee Spector. 2012. Assessment of problem modality by differential performance of lexicase selection in genetic programming: a preliminary report. In Proceedings of the 14th annual conference companion on Genetic and evolutionary computation. 401--408.Google ScholarDigital Library
- Lee Spector and Alan Robinson. 2002. Genetic programming and autoconstructive evolution with the push programming language. Genetic Programming and Evolvable Machines 3 (2002), 7--40.Google ScholarDigital Library
- Peter A Whigham et al. 1995. Grammatically-based genetic programming. In Proceedings of the workshop on genetic programming: from theory to real-world applications, Vol. 16. Citeseer, 33--41.Google Scholar
Index Terms
- On the Trade-Off between Population Size and Number of Generations in GP for Program Synthesis
Recommendations
Population size reduction for the differential evolution algorithm
This paper studies the efficiency of a recently defined population-based direct global optimization method called Differential Evolution with self-adaptive control parameters. The original version uses fixed population size but a method for gradually ...
A generational genetic algorithm with dynamic population size
ACM-SE 47: Proceedings of the 47th Annual Southeast Regional ConferenceIn this paper, a Generational Genetic Algorithm with Dynamic Population Size (GGAWD) is proposed. This algorithm and other Genetic Algorithms (GAs) are applied to F6* and F7* functions in order to find which algorithm is the best. The same evaluation ...
Comments