Using a coevolution mechanism with a Dyna architecture for parameter adaptation in XCS classifier systems

Authors:
Chung-Yuan Huang;Chuen-Tsai Sun;Yen-Wei Chu
Affiliations:
Department of Computer Science and Information Engineering, Yuanpei Institute of Science and Technology, Hsinchu, Taiwan, ROC and Department of Computer Science, National Chiao Tung University, Hs ...;Department of Computer Science, National Chiao Tung University, Hsinchu, Taiwan, ROC;Department of Information Management, Yuanpei Institute of Science and Technology, Hsinchu, Taiwan, ROC and Department of Computer Science, National Chiao Tung University, Hsinchu, Taiwan, ROC
Venue:
CIMMACS'05 Proceedings of the 4th WSEAS international conference on Computational intelligence, man-machine systems and cybernetics
Year:
2005

Citing 5
Cited 0

Adaptation in natural and artificial systems

Adaptation in natural and artificial systems
Introduction to Reinforcement Learning

Introduction to Reinforcement Learning
Classifier fitness based on accuracy

Evolutionary Computation
An analysis of generalization in the xcs classifier system

Evolutionary Computation
Parameter control in evolutionary algorithms

IEEE Transactions on Evolutionary Computation

Quantified Score

Hi-index	0.00

Visualization

Abstract

We propose a co-adaptive approach to control coevolution-based eXtended Classifier System (XCS) parameters. Taking advantage of the on-line incremental learning capabilities of XCS, solutions that completely address target problems can be produced. A coevolution model allows two XCS systems to operate in parallel to solve target and parameter setting problems simultaneously. Since our approach only requires small amounts of information on performance metrics during early run-time stages, it requires little time to become efficient in terms of latent learning. Test results indicate that our proposed system outperforms comparable models regardless of the target problem's stationary/non-stationary status.