Graduation Semester and Year

2012

Language

English

Document Type

Thesis

Degree Name

Master of Science in Electrical Engineering

Department

Electrical Engineering

First Advisor

Michael T Manry

Abstract

A unique algorithm has been developed for training multilayer perceptron neural networks. First the training algorithm has been used with hidden weight optimization (HWO) and multiple optimal learning factors (MOLF) to get the best performance. In each training iteration, this method first optimally orders the inputs and then optimally orders the hidden basis functions. At the end of each training iteration, the method calculates the validation error versus number of basis function curve in one pass through the validation data. Since, pruning is done at each iteration, we optimize validation error over number of basis functions and number of iterations simultaneously. The number of required multiplies for the algorithm has been analyzed. The method has been compared to others in simulations and been found to work very well.

Disciplines

Electrical and Computer Engineering | Engineering

Comments

Degree granted by The University of Texas at Arlington

Download

Share

COinS