Efficient Block Training of Multilayer Perceptrons

The attractive possibility of applying layerwise block training algorithms to multilayer perceptrons MLP, which offers initial advantages in computational effort, is refined in this article by means of introducing a sensitivity correction factor in the formulation. This results in a clear performanc...

Full description

Saved in:
Bibliographic Details
Published in:Neural computation 2000-06, Vol.12 (6), p.1429-1447
Main Authors: Navia-Vázquez, A., Figueiras-Vidal, A. R.
Format: Article
Language:English
Subjects:
Algorithms
Applied sciences
Artificial intelligence
Computer science
control theory
systems
Connectionism. Neural networks
Exact sciences and technology
Expert Systems
Humans
Learning
Neoplasms - pathology
Neural Networks (Computer)
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The attractive possibility of applying layerwise block training algorithms to multilayer perceptrons MLP, which offers initial advantages in computational effort, is refined in this article by means of introducing a sensitivity correction factor in the formulation. This results in a clear performance advantage, which we verify in several applications. The reasons for this advantage are discussed and related to implicit relations with second-order techniques, natural gradient formulations through Fisher's information matrix, and sample selection. Extensions to recurrent networks and other research lines are suggested at the close of the article.
ISSN:0899-7667
1530-888X
DOI:10.1162/089976600300015448