On Simple Adaptive Momentum

Author

Mitchell, R.J.

Author_Institution

Cybern. Intell. Res. Group, Univ. of Reading, Reading

fYear

2008

fDate

9-10 Sept. 2008

Firstpage

1

Lastpage

6

Abstract

Simple Adaptive Momentum was introduced as a simple means of speeding the training of multi-layer perceptrons (MLPs) by changing the momentum term depending on the angle between the current and previous changes in the weights of the MLP. In the original paper, the weight changes of the whole network are used in determining this angle. This paper considers adapting the momentum term using certain subsets of these weights. This idea was inspired by the author´s object oriented approach to programming MLPs, successfully used in teaching students: this approach is also described. It is concluded that the angle is best determined using the weight changes in each layer separately.

Keywords

learning (artificial intelligence); learning systems; multilayer perceptrons; learning systems; multilayer perceptrons training; neural networks; simple adaptive momentum; Backpropagation algorithms; Convergence; Cybernetics; Education; Learning systems; Multi-layer neural network; Multilayer perceptrons; Neural networks; Neurons; Object oriented programming; Back propagation; Learning Systems; Momentum; Neural Networks;

fLanguage

English

Publisher

ieee

Conference_Titel

Cybernetic Intelligent Systems, 2008. CIS 2008. 7th IEEE International Conference on

Conference_Location

London

Print_ISBN

978-1-4244-2914-1

Electronic_ISBN

978-1-4244-2915-8

Type

conf

DOI

10.1109/UKRICIS.2008.4798940

Filename

4798940