DMCA
First and Second-Order Methods for Learning: between Steepest Descent and Newton's Method (1992)
| Venue: | Neural Computation |
| Citations: | 177 - 7 self |
BibTeX
@ARTICLE{Battiti92firstand,
author = {Roberto Battiti},
title = {First and Second-Order Methods for Learning: between Steepest Descent and Newton's Method},
journal = {Neural Computation},
year = {1992},
volume = {4},
pages = {141--166}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
On-line first order backpropagation is sufficiently fast and effective for many large-scale classification problems but for very high precision mappings, batch processing may be the method of choice. This paper reviews first- and second-order optimization methods for learning in feedforward neural networks. The viewpoint is that of optimization: many methods can be cast in the language of optimization techniques, allowing the transfer to neural nets of detailed results about computational complexity and safety procedures to ensure convergence and to avoid numerical problems. The review is not intended to deliver detailed prescriptions for the most appropriate methods in specific applications, but to illustrate the main characteristics of the different methods and their mutual relations.
Keyphrases
second-order method steepest descent computational complexity optimization technique safety procedure different method main characteristic neural net high precision mapping numerical problem mutual relation feedforward neural network many large-scale classification problem appropriate method second-order optimization method specific application detailed result on-line first order backpropagation batch processing many method detailed prescription
