Model selection in neural networks: some difficulties

Curry, Bruce and Morgan, Peter Huw ORCID: https://orcid.org/0000-0002-8555-3493 2006. Model selection in neural networks: some difficulties. European Journal of Operational Research 170 (2) , pp. 567-577. 10.1016/j.ejor.2004.05.026

Full text not available from this repository.

Abstract

This paper considers two related issues regarding feedforward Neural Networks (NNs). The first involves the question of whether the network weights corresponding to the best fitting network are unique. Our empirical tests suggest an answer in the negative, whether using standard Backpropagation algorithm or our preferred direct (non-gradient-based) search procedure. We also offer a theoretical analysis which suggests that there will almost inevitably be functional relationships between network weights. The second issue concerns the use of standard statistical approaches to testing the significance of weights or groups of weights. Treating feedforward NNs as an interesting way to carry out nonlinear regression suggests that statistical tests should be employed. According to our results, however, statistical tests can in practice be indeterminate. It is rather difficult to choose either the number of hidden layers or the number of nodes on this basis.

Item Type:	Article
Date Type:	Publication
Status:	Published
Schools:	Business (Including Economics)
Uncontrolled Keywords:	Neural Networks; Network weights; Hidden layers; Backpropagation; Polytope
Additional Information:	At head of title: Computing, Artificial Intelligence and Information Technology
ISSN:	0377-2217
Last Modified:	17 Oct 2022 09:20
URI:	https://orca.cardiff.ac.uk/id/eprint/2798

Citation Data

Cited 60 times in Scopus. View in Scopus. Powered By Scopus® Data

Actions (repository staff only)

Edit Item