Title :
A neural vector quantizer
Author :
Wang, Zhicheng ; Hanson, John
Author_Institution :
Dept. of Electr. & Comput. Eng., Waterloo Univ., Ont., Canada
Abstract :
A technique of vector quantization based on neural networks called neural vector quantization is investigated. A neural vector quantizer has been developed for data compression and is capable of faster parallel quantization than conventional vector quantizers. The architecture, dynamics, and training strategies are presented. Neural vector quantizers are designed and simulated for Gauss-Markov source data, and the resulting performance, computation, and storage requirements are compared to those for neural vector quantizers of different sizes for the same task and data sources
Keywords :
learning (artificial intelligence); neural nets; parallel algorithms; parallel architectures; vector quantisation; Gauss-Markov source data; architecture; data compression; neural networks; neural vector quantizer; parallel quantization; signal processing; storage requirements; vector quantization; Arithmetic; Computational modeling; Computer architecture; Data compression; Decoding; Encoding; Gaussian processes; Neural networks; Partitioning algorithms; Vector quantization;
Conference_Titel :
Circuits and Systems, 1992. ISCAS '92. Proceedings., 1992 IEEE International Symposium on
Conference_Location :
San Diego, CA
Print_ISBN :
0-7803-0593-0
DOI :
10.1109/ISCAS.1992.229941
