Title :
A multiple-valued bidirectional associative memory
Author_Institution :
Dept. of Inf. Eng., Chinese Univ. of Hong Kong, Shatin, Hong Kong
Abstract :
A multiple-valued asymmetric bidirectional associative memory (MABAM) is proposed. Using multiple-valued neurons, this model can store multiple-valued data pairs, so as to decrease the number of neurons in each layer and hence the number of interconnections. We suggest a learning rule for this model and prove its validity. Simulation results show the performance of this model
Keywords :
content-addressable storage; learning (artificial intelligence); multivalued logic circuits; neural net architecture; interconnections; learning rule validity; multiple-valued asymmetric bidirectional associative memory; multiple-valued data pairs; multiple-valued neurons; performance; simulation; Associative memory; Convergence; Costs; Equations; Error correction; Magnesium compounds; Neural networks; Neurons; Prototypes; Read-write memory;
Conference_Titel :
TENCON '94. IEEE Region 10's Ninth Annual International Conference. Theme: Frontiers of Computer Technology. Proceedings of 1994
Print_ISBN :
0-7803-1862-5
DOI :
10.1109/TENCON.1994.369274
