CS&E Profile: George Karypis

Data mining, bio-informatics, parallel processing, CAD, scientific computing

Ph.D. 1996, B.S. 1992, Computer Science, University of Minnesota

Research

My research interests span the areas of data mining, bio-informatics, parallel processing, CAD, and scientific computing.

My research in data mining is focused on developing innovative new algorithms for a variety of data mining problems including clustering, classification, pattern discovery, and deviation detection, with an emphasis on business applications and information retrieval.

My research in bio-informatics is focused on developing algorithms for understanding the function of genes and proteins in different species using data arising from genome-wide expression profiles. In this work, I'm trying to use data mining techniques to analyze expression profiles of genes and find groups of genes that behave similarly, and determine the underlying genetic regulatory network.

My research in parallel processing is focused on developing scalable parallel algorithms for emerging applications and architectures. This includes research on data intensive applications, scientific computing, architectures with deep memory hierarchies, and architectures with heterogeneous interconnection networks.

My recent research has led to the development of a number of highly efficient and scalable software packages and algorithms such as METIS (a serial sparse graph partitioning software), ParMETIS (an MPI-based parallel graph partitioning software), hMETIS (a circuit partitioning software), PSPASES (a parallel direct solver), and CHAMELEON (a spatial clustering algorithm).