Title :
A modified Hamming neural network with different thresholds and multi-valued weights
Author :
Lu, Wei ; Shi, Bingxue ; Li, Zhijian
Author_Institution :
Inst. of Microelectron., Tsinghua Univ., Beijing, China
Abstract :
A modified Hamming neural network and its integrated circuit model are presented, in which different thresholds and multi-valued weights are introduced. The network consists of a template matching circuit and a “winner-take-all” (WTA) circuit. Simulation shows that it has high precision, high speed and low power dissipation. It can be used to recognize handwritten digits
Keywords :
analogue integrated circuits; circuit analysis computing; neural chips; pattern classification; handwritten digits; high precision; high speed; integrated circuit model; low power dissipation; modified Hamming neural network; multi-valued weights; template matching circuit; thresholds; winner-take-all circuit; Artificial neural networks; Biological neural networks; Biological system modeling; Computational modeling; Hamming distance; Handwriting recognition; Integrated circuit modeling; Neural networks; Pattern matching; Pattern recognition;
Conference_Titel :
Neural Networks, 1996., IEEE International Conference on
Print_ISBN :
0-7803-3210-5
DOI :
10.1109/ICNN.1996.549036
