Title :
Analyzing state space segmentation in learning classifier system
Author :
Wada, Atsushi ; Takadama, Keiki ; Shimohara, Katsunori ; Katai, Osamu
Author_Institution :
ATR, Kyoto, Japan
Abstract :
We present an analysis on state space segmentation for the learning classifier system (LCS). An LCS model is proposed that can segment input state space into variable granularity. A preliminary experiment on a real-valued 6-multiplexor problem is conducted which result revealed that small granularity of segmentation affects the size of the classifier population by causing it to increase.
Keywords :
learning (artificial intelligence); learning systems; pattern classification; state-space methods; classifier population size; learning classifier system; real valued 6-multiplexor problem; state space segmentation; variable granularity;
Conference_Titel :
SICE 2003 Annual Conference
Conference_Location :
Fukui, Japan
Print_ISBN :
0-7803-8352-4
