Title :
Reliability characteristics of Hebbian-type associative memories in network implementation
Author :
Chung, Pau-Choo ; Krile, Thomas F.
Author_Institution :
Dept. of Electr. Eng., Texas Tech Univ., Lubbock, TX, USA
Abstract :
The performance of Hebbian-type associative memories (HAMs) in the presence of faulty interconnections is examined, and equations for predicting network reliability are developed. Optical and VLSI implementations of HAMs are introduced, and the distributions of faulty interconnections in both implementations are discussed. The interconnection faults considered are the equivalent of open-circuit and short-circuit synaptic interconnections. Equations relating the probability of direct one-step convergence (Pdc) to the percentage of failed interconnections are developed for both types of interconnection faults. Monte Carlo simulations indicate that the equations considered can estimate Pdc accurately. Based on the equations, network performance with failed interconnections can be predicted and trade-offs in network design can be determined before proceeding to implementation. The performance of networks with clustered failed interconnections is also discussed and compared with that of networks with randomly distributed faults. The results are discussed from the implementation point of view
Keywords :
Monte Carlo methods; VLSI; content-addressable storage; convergence; failure analysis; integrated memory circuits; neural nets; optical storage; probability; reliability; Hebbian-type associative memories; Monte Carlo simulations; VLSI implementations; clustered failed interconnections; direct on step convergence probability; faulty interconnections; network reliability; open circuit faults; optical implementations; performance; randomly distributed faults; short-circuit synaptic interconnections; Equations; Integrated circuit interconnections; Intelligent networks; Neural networks; Optical computing; Optical fiber networks; Optical interconnections; Optical modulation; Optical sensors; Very large scale integration;
Conference_Titel :
Neural Networks, 1991., IJCNN-91-Seattle International Joint Conference on
Conference_Location :
Seattle, WA
Print_ISBN :
0-7803-0164-1
DOI :
10.1109/IJCNN.1991.155361