Abstract
A key objective of transfer learning is to improve and speed-up learning on a target task after training on a different, but related, source task. This study presents a neuro-evolution method that transfers evolved policies within multi-agent tasks of varying degrees of complexity. The method incorporates behavioral diversity (novelty) search as a means to boost the task performance of transferred policies (multi-agent behaviors). Results indicate that transferred evolved multi-agent behaviors are significantly improved in more complex tasks when adapted using behavioral diversity. Comparatively, behaviors that do not use behavioral diversity to further adapt transferred behaviors, perform relatively poorly in terms of adaptation times and quality of solutions in target tasks. Also, in support of previous work, both policy transfer methods (with and without behavioral diversity adaptation), out-perform behaviors evolved in target tasks without transfer learning.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
Transfer learning and policy transfer are used interchangeably in this paper.
- 2.
All experiments were run in RoboCup Keep-Away version 6 [6]. Source code and executables can be found at: http://people.cs.uct.ac.za/~gnitschke/EvoStar2016/.
- 3.
NEAT and HyperNEAT average maximum task performance progression graphs can be found at: http://people.cs.uct.ac.za/~gnitschke/EvoStar2016/.
References
Pan, S., Yang, Q.: A survey on transfer learning. IEEE Trans. Knowl. Data Eng. 22(10), 1345–1359 (2010)
Torrey, L., Shavlik, J.: Transfer learning. In: Olivas, E.S. (ed.) Handbook of Research on Machine Learning Applications, pp. 17–23. IGI Global, Hershey (2009)
Ammar, H., Tuyls, K., Taylor, M., Driessens, K., Weiss, G.: Reinforcement learning transfer via sparse coding. In: Proceedings of the Eleventh International Conference on Autonomous Agents and Multiagent Systems, Valencia, Spain, pp. 4–8. AAAI (2012)
Ramon, J., Driessens, K., Croonenborghs, T.: Transfer learning in reinforcement learning problems through partial policy recycling. In: Kok, J.N., Koronacki, J., Lopez de Mantaras, R., Matwin, S., Mladenič, D., Skowron, A. (eds.) ECML 2007. LNCS (LNAI), vol. 4701, pp. 699–707. Springer, Heidelberg (2007)
Boutsioukis, G., Partalas, I., Vlahavas, I.: Transfer learning in multi-agent reinforcement learning domains. In: Sanner, S., Hutter, M. (eds.) EWRL 2011. LNCS, vol. 7188, pp. 249–260. Springer, Heidelberg (2012)
Taylor, M., Stone, P., Liu, Y.: Transfer learning via inter-task mappings for temporal difference learning. J. Mach. Learn. 8(1), 2125–2167 (2010)
Stone, P., Kuhlmann, G., Taylor, M.E., Liu, Y.: Keepaway soccer: from machine learning testbed to benchmark. In: Bredenfeld, A., Jacoff, A., Noda, I., Takahashi, Y. (eds.) RoboCup 2005. LNCS (LNAI), vol. 4020, pp. 93–105. Springer, Heidelberg (2006)
Doncleux, S.: Knowledge extraction from learning traces in continuous domains. In: AAAI 2014 Fall Symposium on Knowledge, Skill, and Behavior Transfer in Autonomous Robots, Arlington, USA, pp. 1–8. AAAI Press (2014)
Floreano, D., Dürr, P., Mattiussi, C.: Neuroevolution: from architectures to learning. Evol. Intel. 1(1), 47–62 (2008)
Moshaiov, A., Tal, A.: Family bootstrapping: a genetic transfer learning approach for onsetting the evolution for a set of realated robotic tasks. In: Proceedings of the Congress on Evolutionary Computation, pp. 2801–2808. IEEE Press (2014)
Deb, K.: Pareto Based Multi-objectives Optimization Using Evolutionary Algorithms. Wiley, New York (2001)
Taylor, M., Whiteson, S., Stone, P.: Transfer learning for policy search methods. In: ICML 2006: Proceedings of the Twenty-Third International Conference on Machine Learning Transfer Learning Workshop, Pittsburgh, USA, pp. 1–4. ACM (2006)
Stanley, K., Miikkulainen, R.: Evolving neural networks through augmenting topologies. Evol. Comput. 10(2), 99–127 (2002)
Verbancsics, P., Stanley, K.: Evolving static representations for task transfer. J. Mach. Learn. Res. 11(1), 1737–1763 (2010)
Stanley, K., D’Ambrosio, D., Gauci, J.: A hypercube-based indirect encoding for evolving large-scale neural networks. Artif. Life 15(2), 185–212 (2009)
Stone, P., Sutton, R., Kuhlmann, G.: Reinforcement learning for robocup-soccer keepaway. Adapt. Behav. 13(3), 165–188 (2006)
Whiteson, S., Stone, P.: Evolutionary function approximation for reinforcement learning. J. Mach. Learn. Res. 7(1), 877–917 (2006)
Bahceci, E., Miikkulainen, R.: Transfer of evolved pattern-based heuristics in games. In: Proceedings of the IEEE Symposium on Computational Intelligence and Games, Perth, Australia, pp. 220–227. Morgan Kaufmann (2008)
Lehman, J., Stanley, K.: Abandoning objectives: evolution through the search for novelty alone. Evol. Comput. 19(2), 189–223 (2011)
Mouret, J., Doncieux, S.: Encouraging behavioral diversity in evolutionary robotics: an empirical study. Evol. Comput. 20(1), 91–133 (2012)
Gomes, J., Mariano, P., Christensen, A.: Avoiding convergence in cooperative coevolution with novelty search. In: Proceedings of the International Conference on Autonomous Agents and Multi-agent Systems, pp. 1149–1156. ACM (2014)
Gomes, J., Mariano, P., Christensen, A.: Devising effective novelty search algorithms: a comprehensive empirical study. In: Proceedings of the Genetic Evolutionary Computation Conference, Madrid, Spain, pp. 943–950. ACM (2015)
Degrave, J., Burm, M., Kindermans, P., Dambre, J., Wyffels, F.: Transfer learning of gaits on a quadrupedal robot. Adapt. Behav. 23, 9–19 (2015)
Knudson, M., Tumer, K.: Policy transfer in mobile robots using neuro-evolutionary navigation. In: Proceedings of the Genetic and Evolutionary Computation Conference, Philadelphia, USA, pp. 1411–1412. ACM Press (2012)
Stanley, K.: Compositional pattern producing networks: a novel abstraction of development. Genet. Program Evolvable Mach. 8(2), 131–162 (2007)
D’Ambrosio, D., Stanley, K.: Scalable multiagent learning through indirect encoding of policy geometry. Evol. Intell. J. 6(1), 1–26 (2013)
Gauci, J., Stanley, K.: A case study on the critical role of geometric regularity in machine learning. In: Proceedings of the AAAI Conference on Artificial Intelligence, Menlo Park, USA, pp. 628–633. AAAI Press (2008)
Risi, S., Stanley, K.: Confronting the challenge of learning a flexible neural controller for a diversity of morphologies. In: Proceedings of the Genetic and Evolutionary Computation Conference, pp. 255–261. ACM (2013)
Gomes, J., Christensen, A.: Generic behavior similarity measures for evolutionary swarm robotics. In: Proceedings of the Genetic and Evolutionary Computation Conference, Amsterdam, The Netherlands, pp. 199–206. ACM Press (2013)
Urbano, P., Georgiou, L.: Improving grammatical evolution in santa fe trail using novelty search. In: Proceedings of the 12th European Conference on Artificial Life, Taormina, Italy, pp. 917–924. MIT Press (2013)
Lehman, J., Stanley, K.: Efficiently evolving programs through the search for novelty. In: Proceedings of the 12th Annual Conference on Genetic and Evolutionary Computation, Portland, USA, pp. 837–844. ACM (2010)
Velez, R., Clune, J.: Novelty search creates robots with general skills for exploration. In: Proceedings of the Genetic and Evolutionary Computation Conference, Vancouver, Canada, pp. 737–744. ACM (2014)
Cuccu, G., Gomez, F., Glasmachers, T.: Novelty-based restarts for evolution strategies. In: Proceedings of the Congress on Evolutionary Computation, New Orleans, USA, pp. 158–163. IEEE Press (2011)
Gomes, J., Urbano, P., Christensen, A.L.: Progressive minimal criteria novelty search. In: Pavón, J., Duque-Méndez, N.D., Fuentes-Fernández, R. (eds.) IBERAMIA 2012. LNCS, vol. 7637, pp. 281–290. Springer, Heidelberg (2012)
Lehman, J., Stanley, K.: Revising the evolutionary computation abstraction: minimal criteria novelty search. In: Proceedings of the 12th Annual Conference on Genetic and Evolutionary Computation, pp. 103–110. ACM (2010)
Liapis, A., Yannakakis, G., Togelius, J.: Constrained novelty search: a study on game content generation. Evol. Comput. 23(1), 101–129 (2015)
Ghasemi, A., Zahediasl, S.: Normality tests for statistical analysis: a guide for non-statisticians. Int. J. Endocrinol. Metab. 10(2), 486–489 (2012)
Flannery, B., Teukolsky, S., Vetterling, W.: Numerical Recipes. Cambridge University Press, Cambridge (1986)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2016 Springer International Publishing Switzerland
About this paper
Cite this paper
Didi, S., Nitschke, G. (2016). Multi-agent Behavior-Based Policy Transfer. In: Squillero, G., Burelli, P. (eds) Applications of Evolutionary Computation. EvoApplications 2016. Lecture Notes in Computer Science(), vol 9598. Springer, Cham. https://doi.org/10.1007/978-3-319-31153-1_13
Download citation
DOI: https://doi.org/10.1007/978-3-319-31153-1_13
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-31152-4
Online ISBN: 978-3-319-31153-1
eBook Packages: Computer ScienceComputer Science (R0)