Abstract
Among the models of parallel computing architectures of neural networks, the model of distributed associative memory has very promising features including fault tolerance.
Fault tolerance is here not merely an artificial addition to the existing architecture but intrinsically tied to the basic functions. By addition of a threshold to the linear connection matrix the resulting model is fault-tolerant for errors in input data by providing in one function-cycle a complete pattern recognition process.
The properties of this inherent fault-tolerant process are analytically analyzed and the optimal threshold is calculated.
Furthermore, a hardware model is presented and its fault-tolerance properties are evaluated.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
R. Brause, Fehlertoleranz in intelligenten Benutzerschnittstellen, Informatik Technologie 3, Oldenbourg Verlag 1988, in press
J.A. Feldman, D.H. Ballard Computing with connections, University of Rochester, Computer Science Department, TR72, 1980
K. Fukushima, A Neural Network Model for selective Attention in Visual Pattern Recognition, Biologigal Cybernetics 55, p.5–15, Springer-Verlag 1986
K. Goser, C. Foelster, U. Rueckert, Intelligent memories in VLSI. Information Sciences 34, p61–82, 1984
D. Hillis, The Connection Machine MIT Press, Cambridge, Massachusetts, 1985
Hinton, Anderson, Parallel Models of Associative Memory, Lawrence Erlbaum associates, Hillsdale 1981
Hinton, Implementing Semantic Networks in Parallel Hardware, in /HIN1/
JJ. Hopfield, Neural Networks and physical systems with emergent collective computational abilities, Proc. Natl. Acad. Sci.US A, Vol 79, pp.2554–2558, April 1982
T. Kohonen, Correlation Matrix Memories IEEE Transactions on Computers C21 1972
Kohonen et alii, A demonstration of pattern processing properties of the optimal associative mapping, Froc. Int. Conf. Cybernetics and Society, Washington DC 1977
T. Kohonen, Self-Organisation and Associative Memory, Springer-Verlag Berlin, New York, Tokyo 1984
H.C. Longuet-Higgins, Holographic model of temporal recall, Nature 217, 1968, p. 104
W.S McCulloch, W.H. Pitts, A Logical Calculus of the Ideas Imminent in Neural Nets, Bulletin of Mathematical Biophysics Vol 5, 1943, pp.115–133
D. Willshaw, Models of distributed associative memory, Unpublished doctoral dissertion, Edinburgh University 1971
D. Willshaw, Holography, Association and Induction, in/HINl/
D. Willshaw, O.P. Bunemann, H.C. Longuet-Higgins Non-holographic associative memory Nature, 1969, pp.960–962
Variations on a Theme by Lashley: Lesion Experiments on the Neural Model. Psychological Review 85, 1978
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1988 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Brause, R. (1988). Fault-Tolerance in Non-linear Neural Networks. In: Valk, R. (eds) GI — 18. Jahrestagung II. Informatik-Fachberichte, vol 188. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-74135-7_31
Download citation
DOI: https://doi.org/10.1007/978-3-642-74135-7_31
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-50360-6
Online ISBN: 978-3-642-74135-7
eBook Packages: Springer Book Archive