Fuzzy-XCS: A Michigan genetic fuzzy system

Casillas, Jorge; Carse, Brian; Bull, Larry

Abstract

The issue of finding fuzzy models with an interpretability as good as possible without decreasing the accuracy is one of the main research topics on genetic fuzzy systems. When they are used to perform online reinforcement learning by means of Michigan-style fuzzy rule systems, this issue becomes even more difficult. Indeed, rule generalization (description of state-action relationships with rules as compact as possible) has received a great attention in the nonfuzzy evolutionary learning field (e.g., XCS is the subject of extensive ongoing research). However, the same issue does not appear to have received a similar level of attention in the case of Michigan-style fuzzy rule systems. This may be due to the difficulty in extending the discrete-valued system operation to the continuous case. The intention of this contribution is to propose an approach to properly develop a fuzzy XCS system for single-step reinforcement problems. © 2007 IEEE.

Citation

Casillas, J., Carse, B., & Bull, L. (2007). Fuzzy-XCS: A Michigan genetic fuzzy system. IEEE Transactions on Fuzzy Systems, 15(4), 536-550. https://doi.org/10.1109/TFUZZ.2007.900904

• Journal Article Type: Article
• Publication Date: Aug 1, 2007
• Journal: IEEE Transactions on Fuzzy Systems
• Print ISSN: 1063-6706
• Publisher: Institute of Electrical and Electronics Engineers
• Peer Reviewed: Peer Reviewed
• Volume: 15
• Issue: 4
• Pages: 536-550
• DOI: https://doi.org/10.1109/TFUZZ.2007.900904
• Keywords: Continuous action, genetic fuzzy systems, Michigan-style learning classifier systems, reinforcement learning
• Public URL: https://uwe-repository.worktribe.com/output/1025932
• Publisher URL: http://dx.doi.org/10.1109/TFUZZ.2007.900904
• Additional Information: Additional Information : The main proposal of this paper is to extend the accuracy-based XCS learning classifier system to the fuzzy case, enabling the new fuzzy classifier system to operate using reinforcement learning with continuous valued inputs and outputs. This is a significant contribution since extending the discrete-valued, accuracy-based XCS to the fuzzy domain has long been recognized as a challenging problem. The representation and operators proposed in Fuzzy-XCS are designed to encourage optimal generalization in the evolved fuzzy rule base, providing easier scalability to higher dimensional spaces, faster inference and better linguistic interpretability.