Statistical Applications of Neural Networks

by Sangit Chatterjee and Matthew Laudato.

Abstract: An elementary introduction to the theory of neutral networks is made from a statistical perspective. The networks are estimated by two different methods - one derivative based and the other derivative free. The theory is illustrated by three different examples and where possible results are compared to those obtained from a classical statistical model. The methodology is seen as a new paradigm for data analysis where models are not explicitly stated but rather implicitly defined by the network.

Key Words: Back-propagation, Data Analysis, Feed Forward, Genetic Algorithm.

Authors:
Sangit Chatterjee, chatterje@neu.edu
Matthew Laudato, laudato@neu.edu

Editor: Linda Haines , haines@stat.unp.ac.za

READING THE ARTICLE: You can read the article in portable document (.pdf) format (119615 bytes.)

NOTE: The content of this article is the intellectual property of the authors, who retains all rights to future publication.

This page has been accessed 3105 times since July 24, 2006.


Return to the InterStat Home Page.