Simple Learning Classifier Machine

Abstract

A learning classifier system is one of the methods for applying a genetic-basedapproach to machine learning applications. An enhanced version of the system thatemploys the Bucket-brigade algorithm to reward individuals in a chain of co-operatingrules is implemented and assigned the task of learning rules for classifying simpleobjects. The task is to classify an object that has one or more of the following features:wing, 2-legs/wheels, 3-legs/wheels, 4-legs/wheels, big, flies into one of the following:bird, vehicle. the main goal is to exploit the ability of the algorithm to perform well in anoisy environment and its ability to make little or no assumption about its problemdomain. Results are presented which show that the system was able to learn rules forthe task using only a few training examples and starting with classifiers that wererandomly generated. It is argued that a classifier based learning method requires littletraining examples and that by its use of genetic algorithms to search for new plausiblerules, the method should be able to cope with changing conditions. Results show alsoThe parallel implementation of the algorithm would speed up the training process.