On lookahead and latent learning in simple LCS

ABSTRACT

Learning Classifier Systems use evolutionary algorithms to facilitate rule- discovery, where rule fitness is traditionally payoff based and assigned under a sharing scheme. Most current research has shifted to the use of an accuracy-based scheme where fitness is based on a rule's ability to predict the expected payoff from its use. Learning Classifier Systems which build anticipations of the expected states following their actions are also a focus of current research. This paper presents a simple but effective learning classifier system of this last type, using payoff-based fitness, with the aim of enabling the exploration of their basic principles, i.e., in isolation from the many other mechanisms they usually contain. The system is described and modeled, before being implemented. Comparisons to an equivalent accuracy-based system show similar performance. The use of self-adaptive mutation in such systems in general is then considered.

REFERENCES

• 1: Larry Bull, Lookahead And Latent Learning In ZCS, Proceedings of the Genetic and Evolutionary Computation Conference, p.897-904, July 09-13, 2002
• 2: Bull, L. (2004) Lookahead and Latent Learning in a Simple Accuracy-based Learning Classifier System. In X. Yao et al. (eds) Parallel Problem Solving from Nature - PPSN VIII. Springer Verlag, pp 1042--1050.
• 3: Bull, L. (2005) Two Simple Learning Classifier Systems. In L. Bull & T. Kovacs (eds) Foundations of Learning Classifier Systems. Springer, pp 63--90.
• 4: Bull, L., Hurst, J. & Tomlinson, A. (2000) Self-Adaptive Mutation in Classifier System Controllers. In J -A. Meyer et al. (eds) From Animals to Animats 6 - The Sixth International Conference on the Simulation of Adaptive Behaviour, MIT Press.
• 5: Martin Butz , Wolfgang Stolzmann, An Algorithmic Description of ACS2, Revised Papers from the 4th International Workshop on Advances in Learning Classifier Systems, p.211-230, July 07-08, 2001
• 6: Pierre Gérard , Olivier Sigaud, YACS: Combining Dynamic Programming with Generalization in Classifier Systems, Revised Papers from the Third International Workshop on Advances in Learning Classifier Systems, p.52-69, September 15-16, 2000
• 7: Holland, J. H. (1975) Adaptation in Natural and Artificial Systems. University of Michigan Press. Holland, J. H. (1986) Escaping Brittleness. In R. S. Michalski, J. G. Carbonell & T. M. Mitchell (eds) Machine Learning: An Artificial Intelligence Approach, 2. Morgan Kauffman, pp 48--78.
• 8: J. H. Holland, Concerning the emergence of tag-mediated lookahead in classifier systems, Physica D, v.42 n.1-3, p.188-201, June 1990
• 9: John H. Holland , Keith J. Holyoak , Richard E. Nisbett , Paul R. Thagard, Induction: processes of inference, learning, and discovery, MIT Press, Cambridge, MA, 1986
• 10: Rick L. Riolo, Lookahead planning and latent learning in a classifier system, Proceedings of the first international conference on simulation of adaptive behavior on From animals to animats, p.316-326, February 1991, Paris, France
• 11: Seward, J. P. (1949) An Experimental Analysis of Latent Learning. Journal of Experimental Psychology 39: 177--186.
• 12: Stolzmann, W. (1998) Anticipatory Classifier Systems. In J. R. Koza (ed) Genetic Programming 1998: Proceedings of the Third Annual Conference. Morgan Kaufmann, pp 658--664.
• 13: Wilson, S. W. (1994) ZCS: A Zeroth-level Classifier System. Evolutionary Computation 2(1):1--18.
• 14: Wilson, S. W. (1995) Classifier Fitness Based on Accuracy. Evolutionary Computation 3(2):149--177.

INDEX TERMS

Primary Classification:
  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
      I.2.6 Learning
          Subjects: Knowledge acquisition

General Terms:
Algorithms

Keywords:
anticipation, genetic algorithm, self-adaptation