Title :
A high-dimensional SOFM vector quantizer with weightless neural predictor
Author :
Chen, Yifeng ; Xu, Zhuoqun
Author_Institution :
Dept. of Comput. Sci., Beijing Univ., China
Abstract :
In this paper, a multi-layered compression system is presented based upon a neural vector quantizer called high-dimensional SOFM (HDSOFM). In HDSOFM, neurons are located at vertexes of a hyper-cube. Generally, this algorithm performs better topology-preserving ability. A binary neural predictor and a Huffman encoder are introduced to directly reduce the inter-block redundancy
Keywords :
Huffman codes; image coding; prediction theory; self-organising feature maps; vector quantisation; Huffman encoder; binary neural predictor; high-dimensional SOFM vector quantizer; hyper-cube; inter-block redundancy; multi-layered compression system; neural vector quantizer; topology-preserving ability; weightless neural predictor; Computer science; Data compression; Distortion measurement; Image coding; Network topology; Neural networks; Neurons; Power capacitors; Redundancy; Vector quantization;
Conference_Titel :
Neural Networks, 1995. Proceedings., IEEE International Conference on
Conference_Location :
Perth, WA
Print_ISBN :
0-7803-2768-3
DOI :
10.1109/ICNN.1995.487367
