Abstract

Abstract –Adaptive systems include a vast range of living natural and artificialsystems. Reinforcement learning systems are one form of adaptive systems. The currentwork will focus on a particular kind of reinforcement learning system: the classifiersystem. A classifier system has the ability to categorize its environment and create rulesdynamically, thus making it able to adapt to differing circumstances. This workinvestigates the effect of crowding factor on the classifier system to solve six-bit andeleven-bit multiplexer problems. The six bit multiplexer problem is defined as six signallines that come into the multiplexer. The signals on the first two lines (the address or Alines)are decoded as an assigned binary number. This address value is then used toindicate which of the four remaining signals (on the data or D-lines) is to be passedthrough the multiplexer output. The eleven bit multiplexer problem is defined as elevensignal lines that come into the multiplexer. The signals on the first three lines (theaddress or A-lines) are decoded as an assigned binary number. This address value isthen used to indicate which of the eight remaining signals (on the data or D-lines) is tobe passed through the multiplexer output. This work Investigates the classifier systemrule learning with no crowding and normal crowding settings by comparing andcontrasting the effectiveness of the rule sets learned and their composition in two cases.Experiment results show that the run using classifiers without crowding replacement isunable to perform as well as the run with crowding replacement. The time needed tomatch the signal is shorter when using classifiers with crowding replacement and we aremore likely to achieve good results quickly.

Keywords

The article was added to IASJ on 2014-01-06

223 Total full text downloads since the date of addition