Wednesday 14th June 2000 (6 pm)
ABSTRACT
We develop and study an approach to quantitatively characterizing the inherent information in our data. To do this we use properties of the data which are related to resolution scale and noise. Multiscale entropy of the data is defined, based on the wavelet transform and noise modeling. We demonstrate the effectiveness of such intelligent data modeling. Data semantics, represented by resolution scale and noise properties, are encompassed in our approach. Our goal is to develop general characterization of information content in data, to allow retrieval in unsupervised modes (data mining, clustering) or supervised modes (supervised classification, discriminant analysis).
FIONN MURTAGH (Professor of Computer Science, Queen's University of Belfast; BA, BAI, Trinity College Dublin, 1976; MSc, Information Retrieval, Trinity College Dublin 1979; PhD, Mathematical Statistics, Universite P & M Curie, Paris 6, 1981; Habilitation, University Louis Pasteur, Strasbourg, 1993) has research interests spanning information retrieval, data analysis, and image and signal processing.
He has authored about 10 books including "Image and Data Analysis: the Multiscale Approach", Cambridge University Press, 1998; "Intelligent Information Retrieval", Kluwer, 1993; "Multivariate Data Analysis", Kluwer, 1987; "Multidimensional Clustering Algorithms", Physica-Verlag, 1985; and about 150 journal papers. He is a member of various journal editorial boards and is Editor in Chief of Neurocomputing Letters.
Among 4th Framework projects which he was involved in were ADDSIA (distributed statistical databases) and Neurosat (neural networks for environmental data analysis). He is a partner in the 5th Framework IRAIA project (resource and knowledge discovery in economic databases using intelligent agent and distributed computing technologies) which started in March 2000.
The Evening Lectures are free to both members and non-members of SGES.
For further information contact:
Dr. Hui Liu, Department of Computer Science, Birkbeck College hui@dcs.bbk.ac.uk