An Efficient Method for Extracting Fuzzy Classification Rules from High Dimensional Data

Abstract

We present an efficient method for extracting fuzzy classification rules from high dimensional data. A cluster estimation method called subtractive clustering is used to efficiently extract rules from a high dimensional feature space. A complementary search method can quickly identify the important input features from the resultant high dimensional fuzzy classifier, and thus provides the ability to quickly generate a simpler, more robust fuzzy classifier that uses a minimal number of input features. These methods are illustrated through the benchmark iris data and through two aerospace applications.