CaltechAUTHORS
  A Caltech Library Service

Global descent replaces gradient descent to avoid local minima problem in learning with artificial neural networks

Cetin, Bedri C. and Burdick, Joel W. and Barhen, Jacob (1993) Global descent replaces gradient descent to avoid local minima problem in learning with artificial neural networks. In: 1993 IEEE International Conference on Neural Networks. Vol.2. IEEE , Piscataway, NJ, pp. 836-842. ISBN 0-7803-0999-5. https://resolver.caltech.edu/CaltechAUTHORS:20190612-101006020

[img] PDF - Published Version
See Usage Policy.

574Kb

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20190612-101006020

Abstract

One of the fundamental limitations of artificial neural network learning by gradient descent is the susceptibility to local minima during training. A new approach to learning is presented in which the gradient descent rule in the backpropagation learning algorithm is replaced with a novel global descent formalism. This methodology is based on a global optimization scheme, acronymed TRUST (terminal repeller unconstrained subenergy tunneling), which formulates optimization in terms of the flow of a special deterministic dynamical system. The ability of the new dynamical system to overcome local minima with common benchmark examples and a pattern recognition example is tested. The results demonstrate that the new method does indeed escape encountered local minima, and thus finds the global minimum solution to the specific problems.


Item Type:Book Section
Related URLs:
URLURL TypeDescription
https://doi.org/10.1109/ICNN.1993.298667DOIArticle
Additional Information:© 1993 IEEE.
Record Number:CaltechAUTHORS:20190612-101006020
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20190612-101006020
Official Citation:B. C. Cetin, J. W. Burdick and J. Barhen, "Global descent replaces gradient descent to avoid local minima problem in learning with artificial neural networks," IEEE International Conference on Neural Networks, San Francisco, CA, USA, 1993, pp. 836-842 vol.2. doi: 10.1109/ICNN.1993.298667
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:96317
Collection:CaltechAUTHORS
Deposited By: Tony Diaz
Deposited On:12 Jun 2019 17:14
Last Modified:03 Oct 2019 21:21

Repository Staff Only: item control page