Worst case analysis of weight inaccuracy effects in multilayer perceptrons

Authors
Anguita, D Ridella, S Rovetta, S
Citation
D. Anguita et al., Worst case analysis of weight inaccuracy effects in multilayer perceptrons, IEEE NEURAL, 10(2), 1999, pp. 415-418
Citations number
14
Categorie Soggetti
AI Robotics and Automatic Control
Journal title
IEEE TRANSACTIONS ON NEURAL NETWORKS
ISSN journal
10459227 → ACNP
Volume
10
Issue
2
Year of publication
1999
Pages
415 - 418
Database
ISI
SICI code
1045-9227(199903)10:2<415:WCAOWI>2.0.ZU;2-A
Abstract
We derive here a new method for the analysis of weight quantization effects in multilayer perceptrons based on the application of interval arithmetic, Differently from previous results, we find worst case bounds on the errors due to weight quantization, that are valid for every distribution of the i nput or weight values. Given a trained network, our method allows to easily compute the minimum number of bits needed to encode its weights.