Dan A. Simovici
University of Massachusetts Boston, USA
Compression and data mining
Compression has received lots of attention in the data mining literature. As observed by H. Mannila, data compression can be regarded as one of the fundamental approaches to data mining, since the goal of the data mining is to “compress data by finding some structure in it”.
We examine various compression mechanisms ranging from text-compression, matrix singular decompositions, Fourier-like analysis including wavelets, in analyzing data sets through anomaly
detection, classification and clustering and in evaluating the usefulness of data mining for market basket analysis and graph similarity.
Dan A. Simovici is a Professor of Computer Science and Program Director at the University of Massachusetts Boston and is also affiliated with Dana-Farber Cancer Institute in Boston. Dr. Simovici has received his PhD degree from University of Bucharest. His main research interests are in data mining, machine learning, bioinformatics, soft computing, and multiple-valued logic. He serves as managing editor of the Journal of Multiple-valued Logic and Soft Computing.
Dr. Simovici is the author of more than 160 research papers and of several books. The second edition of his “Mathematical Tools for Data Ming” will appear at Springer in 2014.