UNSUPERVISED NEURAL-NETWORK LEARNING PROCEDURES FOR FEATURE-EXTRACTION AND CLASSIFICATION

In this article, we review unsupervised neural network learning proced ures which can be applied to the task of preprocessing raw data to ext ract useful features for subsequent classification. The learning algor ithms reviewed here are grouped into three sections: information-prese rving methods, density estimation methods, and feature extraction meth ods. Each of these major sections concludes with a discussion of succe ssful applications of the methods to real-world problems.