Title :
Analog parallel matrix structures as tools for neural network investigations
Author :
Rohde, Emil ; Vemanis, Janis
Author_Institution :
Dept. of Electr. Simulation Res., Riga Tech. Univ., Latvia
Abstract :
Parallel resistive structures, developed for physical process modeling, and their possible application in artificial neural systems are considered. The authors presents results of a recent project on hardware for solving boundary field problems and its possible utilization in neurocomputation. The components designed in the project may be useful for experimental neural network engineering as adjustable weights in connections
Keywords :
matrix algebra; neural nets; parallel architectures; analogue parallel matrix structures; boundary field problems; neural network; neurocomputation; parallel resistive structures; Artificial neural networks; Computer networks; Coprocessors; Decoding; Fellows; Hardware; Neural networks; Retina; Silicon; Voltage;
Conference_Titel :
Neuroinformatics and Neurocomputers, 1992., RNNS/IEEE Symposium on
Conference_Location :
Rostov-on-Don
Print_ISBN :
0-7803-0809-3
DOI :
10.1109/RNNS.1992.268562
