Qualitative analysis of neural networks
- 1 February 1989
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Circuits and Systems
- Vol. 36 (2) , 229-243
- https://doi.org/10.1109/31.20200
Abstract
Results from the qualitative theory of large-scale interconnected dynamical systems are surveyed and utilized to develop a qualitative theory for the Hopfield model of neural networks. Such networks are viewed as an interconnection of many single neurons. The results are phrased in terms of the qualitative properties of the individual neurons and in terms of the properties of the interconnecting structure of the neural networks. This method of analysis makes it frequently possible to circumvent difficulties usually encountered in the analysis of complex systems with high dimension. Aspects of neural networks which are addressed include asymptotic stability, exponential stability, and instability of an equilibrium; estimates of trajectory bounds; estimates of the domain of attraction of an asymptotically stable equilibrium: and stability of neural networks under structural perturbations (arising, e.g. during adaptive learning schemes). The results are not overly conservative. Furthermore, they are in a form which will make them highly useful as constraints in synthesis or design procedures.<>Keywords
This publication has 13 references indexed in Scilit:
- VISUALIZING LEARNING AND COMPUTATION IN ARTIFICIAL NEURAL NETWORKSInternational Journal on Artificial Intelligence Tools, 1992
- Qualitative analysis and synthesis of a class of neural networksIEEE Transactions on Circuits and Systems, 1988
- Neural networks for nonlinear programmingIEEE Transactions on Circuits and Systems, 1988
- Computing with Neural Circuits: A ModelScience, 1986
- Simple 'neural' optimization networks: An A/D converter, signal decision circuit, and a linear programming circuitIEEE Transactions on Circuits and Systems, 1986
- Neurons with graded response have collective computational properties like those of two-state neurons.Proceedings of the National Academy of Sciences, 1984
- On the status of stability of interconnected systemsIEEE Transactions on Automatic Control, 1983
- Stability analysis of complex dynamical systemsCircuits, Systems, and Signal Processing, 1982
- Neural networks and physical systems with emergent collective computational abilities.Proceedings of the National Academy of Sciences, 1982
- Some Theorems on the Dynamic Response of Nonlinear Transistor NetworksBell System Technical Journal, 1969