8:30 - 11:45 AM, 18 May, 2002
Blind source separation is a field that has garnered much recent research and commercial interest in fields as diverse as digital and wireless communications, acoustics, and medicine. This tutorial provides an overview of the theoretical criteria, useful algorithms, and practical applications of blind source separation to signal processing practitioners and researchers. An emphasis is placed on understanding the capabilities of and relationships between various criteria and methods so that their usefulness for a particular practical task can be gauged. Topics for the tutorial include:
1. Definitions of and relationships between blind source separation, independent component analysis, blind deconvolution/equalization, and convolutive blind source separation;
2. Information-theoretic, higher-order statistics-based, and second-order statistics-based criteria for blind source separation;
3. Single- vs. multi-stage separation structures;
4. Novel algorithmic concepts such as the natural gradient, the relative gradient, efficient adaptation of orthonormal systems, and equivariance;
5. Whitening constraints in blind source separation;
6. Time- vs. frequency-domain implementations of convolutive blind source separation methods;
7. Blind minimum mean-squared error source separation in noisy environments; and
8. Applications to narrowband array processing in communications and wideband speech separation for multiple-microphone audio recordings.
The assumed background for the tutorial attendee is a bachelor of science degree in an engineering or scientific discipline along with a general knowledge of digital and adaptive signal processing systems and methods.
presenter:
Scott C. Douglas is an associate professor in the Department of Electrical Engineering at Southern Methodist University, Dallas, Texas. He received his B.S., M.S., and Ph.D. degrees from Stanford University, Stanford, California. Dr. Douglas' research efforts span the fields of adaptive filters, blind source separation, and active noise control. He has authored or co-authored six book chapters and over 100 refereed journal and conference papers in these fields. He was the recipient of an NSF CAREER Young Investigator Award in 1995, and he has received research funding from the U.S. Army and other U.S. governmental organizations, the State of Texas, and Raytheon E-Systems Company. He is currently Chair of the Neural Networks for Signal Processing Technical Committee and Secretary of the Signal Processing Education Technical Committee of the IEEE Signal Processing Society. He has been an Associate Editor for both the IEEE Transactions on Signal Processing and IEEE Signal Processing Letters. Dr. Douglas is regularly involved in organizing professional meetings, most notably as the Proceedings Editor of the 1999 International Symposium on Active Control of Sound and Vibration and the Proceedings Co-Editor of the 1999, 2000, 2001, and 2002 IEEE Workshops on Neural Networks for Signal Processing. Most recently, he has designed training materials and co-authored a textbook, entitled Multimedia and Information Engineering (Prentice Hall, 2002), for the INFINITY Project, a multi-faceted effort to establish a nationwide engineering curriculum at the high school level. Dr. Douglas is a frequent consultant to industry, a senior member of the IEEE, and a member of both Phi Beta Kappa and Tau Beta Pi.