Improving a Pittsburgh learnt fuzzy rule base using feature subset selection

This paper investigates the problem of feature subset selection as a preprocessing step to a method which learns fuzzy rule bases using genetic algorithm (GA) implementing the Pittsburgh approach. Four feature subset selection methods are investigated in the context of learning fuzzy rule bases. Two of them are filter methods namely, the Relief-E and the C-Focus. The other two are wrapper methods using GA as their search process; one implements the instance-based method 1-NN and the other, the constructive neural network algorithm DistAI. Results of the experiments conducted in three domains are presented and discussed; they show that methods which learn fuzzy rule bases can benefit from feature subset selection methods.