Neural Networks Based Feature Selection in Biological Data Analysis

Abstract

In this chapter we present a novel method for scoring function specification and feature selection by combining unsupervised learning with supervised cross validation. Various clustering algorithms such as one dimensional Kohonen SOM, k-means, fuzzy c-means and hierarchical clustering procedures are used to perform a clustering of object-data for a chosen subset of input features and a given number of clusters. The resulting object clusters are compared with the predefined target classes and a matching factor (score) is calculated. This score is used as criterion function for heuristic sequential and cross feature selection.