ABSTRACT
8.1 Introduction
So far we have described various data mining tasks such as condensation, feature selection, learning, case generation and clustering. The present chapter deals with the tasks of classification, rule generation and rule evaluation. Here we describe a synergistic integration of four soft computing components, namely, fuzzy sets, rough sets, neural networks and genetic algorithms along with modular decomposition strategy, for generating a modular rough-fuzzy multilayer perceptron (MLP). The resulting connectionist system achieves gain in terms of performance, learning time and network compactness for classification and linguistic rule generation. Different quantitative indices are used for evaluating the linguistic rules, and to reflect the knowledge discovery aspect.
