Application of Artificial Neural Networks for Real Time Data Compression
Data compression techniques often involve compact representations by identifying and using structures inherent within the data itself. These structures usually have some degree of variance between themselves. An Artificial Neural Network (ANN) approach helps simplify the recognition of these complex variances and subsequently the classification of these data. Real time data compression involves the identification of the Probability Distribution Function (pdfs) or the density of each of the features for each file containing such data. By using this ANN approach for data classification, one can then encode a Huffman data compression engine. If this scheme could be employed on an Internet Service Provider (ISP) server, then at the user end a decoder (a Huffman decoder engine) could retrieve the original values. The objective of this paper is to demonstrate a scheme for compressing data based on the feature content of the data present within html files, which make up the bulk of all Internet related formats. It is based on some experimental work involving SelfOrganizing Maps and a Huffman encoder/decoder engine.
Winston DSouza Tim Spracklen
Department of Engineering University of Aberdeen Aberdeen - AB24 3UE Scotland, UK
国际会议
8th International Conference on Neural Information Processing(ICONIP 2001)(第八届国际神经信息处理大会)
上海
英文
727-730
2001-11-14（万方平台首次上网日期，不代表论文的发表时间）