Date of Award

Fall 12-1-2004

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

College of Technology

First Advisor

W. Tad Foster

Second Advisor

David Beach

Third Advisor

Douglas Herrmann

Abstract

It is believed that a distributed network system will provide a general means of understanding cognitive functions that occur in the brain. This study provides a framework for modeling human memory behavior using a Distributed Artificial Neural Network (DANN) architecture. This approach presents a different method for modeling cognitive functions using peer-to-peer computing. The computers on the network represents different memory stores, and by using Hopfield Neural Network Techniques, along with the software application Joone, the final product is a DANN system that exhibits human memory behavior and the ability to associate input vectors from the Fisher's Iris Classification output data. The human memory aspect of the study is validated using predefine data sets and tests that compromises the methodology. These tests are also used to demonstrate the DANN's ability to use associative memory, similar to the Hopfield Neural Network. The memory behavior was based on the theoretical aspects of human memory proposed by Atkinson and Shiffrin in 1968, although there are several researchers that do not support the Atkinson and Shiffrin Human Memory Model, the simplicity of their model is still being used as the basic behavior of memory.

Download

Share

COinS