Dissertation/Thesis Abstract

An exploration of varying conditions in a Hopfield Neural Network and applications to a DNA implementation
by Hughes, Bradley Steven, Ph.D., University of California, Riverside, 2010, 233; 3426183
Abstract (Summary)

A Hopfield Neural Network is a content addressable memory with elements consisting of the correlations between elements of memory vectors. Recall of a complete memory vector is possible via the introduction of a “corrupted” vector, which is a memory vector with some components altered. It may also be possible to correctly recall memories with the use of a partial vector. It may be possible to create such an information storage and retrieval system using DNA as a working substance. Herein I present some computational results for properties of Hopfield Neural Networks, as well as a theoretical framework for the operation of such a system, including possible limitations in the working substance.

Indexing (document details)
Advisor: Mills, Allen P., Jr.
Commitee: Beyermann, Ward, Zandi, Roya
School: University of California, Riverside
Department: Physics
School Location: United States -- California
Source: DAI-B 71/12, Dissertation Abstracts International
Subjects: Neurosciences, Biophysics
Keywords: DNA implementation, Hopfield, Memory, Neural network
Publication Number: 3426183
ISBN: 978-1-124-26372-4
Copyright © 2020 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy