Neighborhood based Levenberg-Marquardt algorithm for neural network training

Although the Levenberg-Marquardt (LM) algorithm has been extensively applied as a neural-network training method, it suffers from being very expensive, both in memory and number of operations required, when the network to be trained has a significant number of adaptive weights. In this paper, the be...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transaction on neural networks and learning systems 2002-09, Vol.13 (5), p.1200-1203
Main Authors: Lera, G., Pinzolas, M.
Format: Article
Language:English
Subjects:
Adaptive algorithms
Algorithms
Backpropagation algorithms
Computation
Computer languages
Iterative methods
Multilayer perceptrons
Networks
Neural networks
Neurons
Occupation
Optimization methods
Software performance
Software tools
Testing
Training
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:Although the Levenberg-Marquardt (LM) algorithm has been extensively applied as a neural-network training method, it suffers from being very expensive, both in memory and number of operations required, when the network to be trained has a significant number of adaptive weights. In this paper, the behavior of a recently proposed variation of this algorithm is studied. This new method is based on the application of the concept of neural neighborhoods to the LM algorithm. It is shown that, by performing an LM step on a single neighborhood at each training iteration, not only significant savings in memory occupation and computing effort are obtained, but also, the overall performance of the LM method can be increased.
ISSN:1045-9227
2162-237X
1941-0093
2162-2388
DOI:10.1109/TNN.2002.1031951