For Full-Text PDF, please login, if you are a member of IEICE,|
or go to Pay Per View on menu list, if you are a nonmember of IEICE.
Optimal Design of Hopfield-Type Associative Memory by Adaptive Stability-Growth Method
Xue-Bin LIANG Toru YAMAGUCHI
IEICE TRANSACTIONS on Information and Systems
Publication Date: 1998/01/25
Print ISSN: 0916-8532
Type of Manuscript: LETTER
Category: Bio-Cybernetics and Neurocomputing
associative memory, neural network, optimal stability, adaptive stability-growth method,
Full Text: PDF(204.5KB)>>
An adaptive stability-growth (ASG) learning algorithm is proposed for improving, as much as possible, the stability of a Hopfield-type associative memory. While the ASG algorithm can be used to determine the optimal stability instead of the well-known minimum-overlap (MO) learning algorithm with sufficiently large lower bound for MO value, it converges much more quickly than the MO algorithm in real implementation. Therefore, the proposed ASG algorithm is more suitable than the MO algorithm for real-world design of an optimal Hopfield-type associative memory.