Title :
Spatio-temporal pattern recognition by competitive networks
Author :
Tanaka, Takehisa
Author_Institution :
Matsushita Res. Inst. Tokyo Inc., Kawasaki, Japan
Abstract :
Proposes a spatio-temporal pattern recognition model. The network transforms temporally-continuous spatially-distributed patterns into temporally-segmented spatially-local representation. The model is based on a competitive network models proposed by Amari (1978) and Arbib (1989) but the author´s network structure and parameters are different from theirs and the biggest-input-taker always wins with adjustable hysteresis.
Keywords :
neural nets; pattern recognition; adjustable hysteresis; competitive networks; spatio-temporal pattern recognition; temporally-continuous spatially-distributed patterns; temporally-segmented spatially-local representation; Character recognition; Fires; Hysteresis; Keyboards; Mathematical model; Neural networks; Optical character recognition software; Optical network units; Pattern recognition; Speech recognition;
Conference_Titel :
Neural Networks, 1993. IJCNN '93-Nagoya. Proceedings of 1993 International Joint Conference on
Print_ISBN :
0-7803-1421-2
DOI :
10.1109/IJCNN.1993.716809
