Friday, April 21, 2000 B102 Wells Hall:

• 4:00 pm     Ingrid Daubechies
                   "From Nonlinear Approximation Theorems to Rate-Distortion Bounds"
• Abstract: Because wavelets are unconditional bases for many function spaces, they give rise to good nonlinear approximation bounds. These have been viewed by mathematicians as the "explanation" of why wavelets work well for compression in a variety of settings. Yet the practice of compression is different from the question of how well we can approximate in a given basis if we are allowed to keep any N terms we choose - in practice, one has to measure the (expected) distortion as a function of the total number of bits used to convey the information. It turns out that in order to exploit the nonlienar approximation results, one has to combine them with "smart" coding strategies (invented already by the engineers!). This is joint work with Albert Cohen, Onur Guleryuz and Michael Orchard.

For additional information:

Department of Mathematics
Michigan State University
East Lansing, MI 48824-1027
(517)355-9680
ginther@math.msu.edu
http://www.math.msu.edu/Lecture_Series