Peter J. Edwards and Alan F. Murray Dept. of Electrical Engineering Edinburgh ...  Home/Search
Context Related
| ee.ed.ac.uk/~pje/pub...pje_ICNN96.ps.gz Cached: PS.gz PS PDF This document uses CoBlitz to cache paper downloads. If your firewall is blocking outgoing connections to port 3125, you can use these links to download local copies.
Image Update HelpPS.gz PS PDF From: ee.ed.ac.uk/~pje/publications.... (more) (Enter author homepages) |
Rate this article:      (best)Comment on this article |
Abstract: This paper presents a study of weight- and input-noise in feed-forward network training algorithms. In theory for the optimal least-squares case noise can be modelled by a single cost function term. However we believe that such ideal conditions are uncommon in practice. Both first and second derivative terms are shown to have the potential to de-sensitize the trained network's outputs to weight- or input-corruption. Simulation experiments illustrate these points comparing the ideal case with a... (Update)
Context of citations to this paper: More
.... the learning process and we have investigated the conditions under which weight noise is equivalent to including cost function penalties [4, 6]. The penalties were considered to be constraints on weight saliency. Here we expand this idea onto that of smoothness of the solution...
...learning convergence. Work has also been carried out to ascertain the underlying mechanisms of the performance improvements[9] 2] [10]. The use of weight noise has therefore been shown to be beneficial for a number of applications all trained and tested in software...
Cited by: More
Fault-tolerance via weight-noise in analogue VLSI.. - Edwards, Murray (1997) (Correct)
Penalty Terms for Fault Tolerance - Edwards, Murray (1997) (Correct)
Towards Optimally Distributed Computation - Edwards, Murray (1997) (Correct)
Active bibliography (related documents): More All
0.5: Extraction of Rules from Discrete-Time Recurrent Neural Networks - Omlin, Giles (1996) (Correct)
0.1: Can Deterministic Penalty Terms Model the Effects of Synaptic .. - Edwards, Murray (1995) (Correct)
0.1: Towards Optimally Distributed Computation in Augmented Networks - Edwards, Murray (Correct)
Similar documents based on text: More All
0.1: Committee Formation for Reliable and Accurate Neural.. - Edwards, Murray (2000) (Correct)
0.0: Enhanced MLP Performance and Fault Tolerance Resulting from.. - Murray, Edwards (1994) (Correct)
0.0: Analogue Synaptic Noise - Implications and Learning Improvements - Edwards, Murray (1993) (Correct)
Related documents from co-citation: More All
3: Neural networks for admission control in an ATM network (context) - Nordstrom, Gallmo et al. - 1992
3: Synaptic weight noise during MLP training : Enhanced MLP performance and fault t.. (context) - Murray, Edwards - 1994
3: Hardware learning in analogue VLSI neural networks - Lehmann - 1994
BibTeX entry: (Update)
Edwards, P., and Murray, A. 1996b. Modelling weight- and input-noise in MLP learning. Proc. http://citeseer.ist.psu.edu/84186.html More
@misc{ edwards-modelling,
author = "P. Edwards and A. Murray",
title = "Modelling weight- and input-noise in MLP learning",
text = "Edwards, P., and Murray, A. 1996b. Modelling weight- and input-noise in
MLP learning. Proc.",
url = "citeseer.ist.psu.edu/84186.html" }
Citations (may not include all citations):1535 Cambridge University Press (context) - Press, Teukolsky et al. - 1992
113 Analysis of hidden units in a layered network trained to cla.. (context) - Gorman, Sejnowski - 1988
81 Neural networks and related methods for classification (context) - Ripley - 1994
57 Training with noise is equivalent to Tikhonov regularization - Bishop - 1995
50 The multilayer perceptron as an approximation to a bayes opt.. (context) - Ruck, Rogers et al. - 1990
7 Synaptic weight noise during MLP training : Enhanced MLP per.. (context) - Murray, Edwards - 1994
4 Can deterministic penalty terms model the effects of synapti.. - Edwards, Murray - 1996
2 Effects of noise on convergence and generalization in recurr.. - Jim, Horne et al. - 1995
1 From data distributions to regularisation in invarient learn.. (context) - Leen - 1995
Documents on the same site (http://www.ee.ed.ac.uk/~pje/publications.html/): More
Synaptic Weight Noise During MLP Learning Enhances.. - Murray, Edwards (1993) (Correct)
Weight Saliency Regularisation in Augmented Networks - Peter Edwards (Correct)
Towards Optimally Distributed Computation - Edwards, Murray (1997) (Correct)
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC