Vladimir S. Cherkassky
Associate Professor

M.S., 1976, Sys. Eng. and Operations Research, Moscow Aviation Institute
Ph.D., 1985, Elect. and Computer Eng., University of Texas at Austin

Telephone: (612) 625-9597
Office Hours

E-mail: cherkass@ece.umn.edu
Web Page: http://www.ece.umn.edu/users/cherkass/


My research interests include parallel processing, computer networks, fault-tolerant computing and neural networks. My current efforts in the area of parallel and fault-tolerant computing are concerned with effective utilization of built-in redundancy of multicomputer systems to enhance system's reliability and to achieve graceful degradation. I am also interested in real-time local area network (LAN) architectures, and in particular in evaluation of various priority handling methods in LANs provided by existing standards.

A major part of my current research is focussed on Neural Networks. I am interested in scaling and computational properties of neural network models, and in neural network applications to associative database retrieval, character recognition and expert systems.


Selected Publications

"Learning from Data: Concepts, Theory and Methods", V. Cherkassky and F. Mulier, Wiley Interscience, 1998.

"From Statistics To Neural Networks. Theory and Pattern Recognition Applications", V. Cherkassky , J.H. Friedman and H. Wechsler (Eds.), NATO ASI Series F, v.136, Springer-Verlag, 1994.

"Model selection for regression using VC generalization bounds", V. Cherkassky, X. Shao, F. Mulier and V. Vapnik, IEEE Trans on Neural Networks, 10,5, 1999, 1075-1089

"Measuring the VC-dimension using optimized experimental design," X. Shao, V. Cherkassky and W. Li, Neural Computation, MIT Press, 2000, 12, 8, 1969-1986

"Signal estimation and denoising using VC-theory", Cherkassky and X. Shao, Neural Networks, Pergamon, 14, 2001, 37-52

"Model complexity control and statistical learning theory", V. Cherkassky, Natural Computing, Kluwer,1,2002, 109-133.