Title :
Quantization effects of Hebbian-type associative memories
Author :
Chung, Pau-Choo ; Chung, Pau-Choo ; Tsai, Ching-Tsorng
Author_Institution :
Dept. of Electr. Eng., Nat. Cheng Kung Univ., Tainan, Taiwan
Abstract :
The effects of quantization strategies in Hebbian-type associative memories are explored. The quantization strategies considered include two-level, three-level with a cut-off threshold, and linear quantizations. The two-level strategy is to clip positive interconnections into +1 and negative interconnections into -1. The three-level quantization uses the same strategy in turning the interconnections into +1 or -1, except that it is applied only to those interconnections having their values larger than a cut-off threshold. Those interconnections within the cutoff threshold are then set to zero. The results indicate that three-level quantization with a properly selected cut-off threshold gives higher network performance than two-level quantization. The performance of a network with linear quantization is also compared with a network with three-level quantization. It is found that the linear quantization does not significantly enhance network performance compared with the three-level quantization
Keywords :
Hebbian learning; content-addressable storage; neural nets; Hebbian-type associative memories; cut-off threshold; linear quantizations; negative interconnections; network performance; positive interconnections; quantization strategies; three-level; two-level; Associative memory; Computer networks; Equations; Hardware; Hebbian theory; Neurodynamics; Quantization; Turning; Very large scale integration;
Conference_Titel :
Neural Networks, 1993., IEEE International Conference on
Conference_Location :
San Francisco, CA
Print_ISBN :
0-7803-0999-5
DOI :
10.1109/ICNN.1993.298756