Date of Award
2002-5
Degree Name
Master of Science
Department
Department of Mathematical Sciences
Advisor(s)
Patricia Nava
Second Advisor
William Kaigh
Third Advisor
Gavin Gregory
Abstract
Neural Networks (NN) have been used successfully to solve many different problems. It is an area of computer science and statistics that is very useful in real life. The advantages of NN consist of their ability to produce nonlinear input-output mapping, their adaptability, fault tolerance, uniform analysis and design. The search for better ways to train them consists of trying to find the optimal values of their free parameters. Some of these training algorithms are based on Markov Chains. They have some advantages over the traditionally used deterministic methods. In this document the main ideas related to these concepts are presented. The results of these simulations are also included.
Language
en
Rights Holder
Raul Cruz-Cano
Recommended Citation
Cruz-Cano, Raul, "On Markov Chain Monte Carlo Methods for Neural Networks" (2002). Open Access Theses & Dissertations. 4503.
https://scholarworks.utep.edu/open_etd/4503