Title :
Novel synthesis procedure for an asynchronous discrete-time associative memory
Author :
Araki, Kenji ; Nowara, Kenji ; Saito, Toshimichi
Author_Institution :
Dept. of Electr. Eng., Hosei Univ., Tokyo, Japan
Abstract :
A novel synthesis procedure for an asynchronous discrete-time neural network is proposed. It uses a weighted correlation matrix, and its algorithm includes a linear equation of weighting coefficients. If all desired memories are independent, the synthesis procedure guarantees the storage of all desired memories and that there are no oscillatory sequences. Experiments show that the procedure exhibits much better storage performance than previous ones
Keywords :
content-addressable storage; discrete time systems; neural nets; asynchronous discrete-time neural network; linear equation; oscillatory sequences; storage performance; synthesis procedure; weighted correlation matrix; weighting coefficients; Associative memory; Symmetric matrices;
Conference_Titel :
Circuits and Systems, 1992. ISCAS '92. Proceedings., 1992 IEEE International Symposium on
Conference_Location :
San Diego, CA
Print_ISBN :
0-7803-0593-0
DOI :
10.1109/ISCAS.1992.229912