CIRCUIT THEORETIC SOLUTIONS FOR NEURAL NETWORKS - AN OLD APPROACH TO A NEW PROBLEM.
Abstract
A major hindrance to research in the area of neural systems is the lack of adequate analytical tools for the study of the nonlinear phenomena (notably multiple equilibria and complex dynamics) manifested by neural networks. The field of nonlinear circuit analysis is rich with tools of the type necessary to explain many of these phenomena. The authors explore the use of one such tool, the cocontent function, illustrating its use in the stability analysis of Hopfield-type neural networks. They show how the function can be used in the analysis of content-addressable memories and to produce a reliable analog nonlinear programming circuit based on a neural-network scheme.
Recommended Citation
Kennedy, M., & Chua, L. (1987). CIRCUIT THEORETIC SOLUTIONS FOR NEURAL NETWORKS - AN OLD APPROACH TO A NEW PROBLEM.. Retrieved from https://digitalcommons.memphis.edu/facpubs/17364
