DocumentCode :
970181
Title :
Function Approximation With XCS: Hyperellipsoidal Conditions, Recursive Least Squares, and Compaction
Author :
Butz, Martin V. ; Lanzi, Pier Luca ; Wilson, Stewart W.
Author_Institution :
Dept. of Psychol., Univ. of Wurzburg, Wurzburg
Volume :
12
Issue :
3
fYear :
2008
fDate :
6/1/2008 12:00:00 AM
Firstpage :
355
Lastpage :
376
Abstract :
An important strength of learning classifier systems (LCSs) lies in the combination of genetic optimization techniques with gradient-based approximation techniques. The chosen approximation technique develops locally optimal approximations, such as accurate classification estimates, Q-value predictions, or linear function approximations. The genetic optimization technique is designed to distribute these local approximations efficiently over the problem space. Together, the two components develop a distributed, locally optimized problem solution in the form of a population of expert rules, often called classifiers. In function approximation problems, the XCSF classifier system develops a problem solution in the form of overlapping, piecewise linear approximations. This paper shows that XCSF performance on function approximation problems additively benefits from: 1) improved representations; 2) improved genetic operators; and 3) improved approximation techniques. Additionally, this paper introduces a novel closest classifier matching mechanism for the efficient compaction of XCS´s final problem solution. The resulting compaction mechanism can boil the population size down by 90% on average, while decreasing prediction accuracy only marginally. Performance evaluations show that the additional mechanisms enable XCSF to reliably, accurately, and compactly approximate even seven dimensional functions. Performance comparisons with other, heuristic function approximation techniques show that XCSF yields competitive or even superior noise-robust performance.
Keywords :
function approximation; genetic algorithms; gradient methods; learning (artificial intelligence); least squares approximations; pattern classification; pattern matching; piecewise linear techniques; recursive functions; XCSF classifier system; closest classifier matching mechanism; compaction, condensation; function approximation; genetic optimization technique; gradient-based approximation technique; hyperellipsoidal condition; learning classifier system; piecewise linear approximation; recursive least square; Closest classifier matching (CCM); XCS; compaction; condensation; function approximation; genetic algorithm (GA); hyperellipsoids; learning classifier system (LCS); neural network (NN); recursive least squares (RLS); self organization;
fLanguage :
English
Journal_Title :
Evolutionary Computation, IEEE Transactions on
Publisher :
ieee
ISSN :
1089-778X
Type :
jour
DOI :
10.1109/TEVC.2007.903551
Filename :
4380293
Link To Document :
بازگشت