Date of Award

Fall 12-1-2004

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

Electrical Engineering

First Advisor

David Beach

Second Advisor

Tad Foster

Third Advisor

Douglas Herrmann

Abstract

It is difficult to be certain about brain functioning. Until now, most efforts to understand brain computations have only been conducted on a stand alone computer. In order to recreate the brain accurately, members of the Institute of Cognitive Computing Technology (ICCT) were interested in finding out whether or not a biological process could be simulated on a Distributed Artificial Neural Network (DANN). The goal of this investigation was to implement visual perception on a network of computers and compare the computations of the network with those of a human brain. The distributed network was developed as a Local Area Network in the ICCT lab at Indiana State University using three computers and a Cisco switch. The stimulus selected to teach and test the network was a modified version of Fisher's iris data measured with the vision system. In this study, only sepal length and width were used as inputs. Using a backpropagation structure, the network was trained and validated in a distributed environment. The network performance was then compared to that of a human brain, revealing similarities between the two.

Share

COinS