Empirical evidence for ultrametric structure in multi-layer perceptron error surfaces

Gallagher, Marcus, Downs, Tom and Wood, Ian (2002) Empirical evidence for ultrametric structure in multi-layer perceptron error surfaces. Neural Processing Letters, 16 2: 177-186. doi:10.1023/A:1019956303894


Author Gallagher, Marcus
Downs, Tom
Wood, Ian
Title Empirical evidence for ultrametric structure in multi-layer perceptron error surfaces
Journal name Neural Processing Letters   Check publisher's open access policy
ISSN 1370-4621
Publication date 2002-10-01
Year available 2002
Sub-type Article (original research)
DOI 10.1023/A:1019956303894
Open Access Status Not yet assessed
Volume 16
Issue 2
Start page 177
End page 186
Total pages 10
Place of publication The Netherlands
Publisher Kluwer Academic Publishers
Language eng
Subject C1
280212 Neural Networks, Genetic Alogrithms and Fuzzy Logic
780101 Mathematical sciences
280200 Artificial Intelligence and Signal and Image Processing
Abstract Combinatorial optimization problems share an interesting property with spin glass systems in that their state spaces can exhibit ultrametric structure. We use sampling methods to analyse the error surfaces of feedforward multi-layer perceptron neural networks learning encoder problems. The third order statistics of these points of attraction are examined and found to be arranged in a highly ultrametric way. This is a unique result for a finite, continuous parameter space. The implications of this result are discussed.
Keyword Computer Science, Artificial Intelligence
Neurosciences
Configuration Space Analysis
Error Surface
Feedforward Neural Network
Multi-layer Perceptron
Ultrametricity
Neural-network
Q-Index Code C1
Institutional Status UQ

 
Versions
Version Filter Type
Citation counts: TR Web of Science Citation Count  Cited 1 times in Thomson Reuters Web of Science Article | Citations
Scopus Citation Count Cited 1 times in Scopus Article | Citations
Google Scholar Search Google Scholar
Created: Wed, 15 Aug 2007, 03:46:13 EST