Special Seminars (with Room Correction)

Robert M. Gray gray at ee.Stanford.EDU
Mon Dec 2 10:34:25 PST 2002


Please note both talks are in the Packard EE Building Auditorium,
Packard 101.


Dr. Jelena Kovacevic of Bell Labs (Lucent) will be visiting Stanford
December 3- December 5 and will be giving two special EE Seminars during
her visit. The Wednesday talk is intended for a more general audience.


     	    Life After Wavelets: The War Of The Frames
		     
		    Jelena Kovacevic, Bell Labs

             Tuesday, December 3rd Noon Packard 101

		    
			  ABSTRACT

The addition of wavelets to the signal processing toolbox greatly
enhanced our ability to deal with nonstationary signals. Its effects
are felt beyond the research community through JPEG, now wavelet
based.

Frames -- redundant representations of which wavelets are a subclass
-- are the next addition to the toolbox. Tight frames, in particular,
with a few restrictions, become orthonormal bases. I talk here about
finite tight frames, fundamental for a broad spectrum of applications.
Until now it was thought that such frames were sparse.  I show that
actually tight frames are everywhere and they can be custom built for
most applications as long as the requirements are not too rigid and do
not violate the Fundamental Inequality which all tight frames must
satisfy. This inequality governs the distribution of power among the
frame vectors. In this ``war of the frames'', if the Fundamental
Inequality is not satisfied, the dominant vectors grab dimensions to
themselves and let the rest of the vectors squabble over the remaining
dimensions. On the other hand, if the Fundamental Inequality holds, no
vector is powerful enough to subjugate the rest, leading to equitable
power sharing.


			Photo-to-Grandma Problem:
			Compression Meets the Network

			Jelena Kovacevic, Bell Labs

                  Wednesday, December 4th  4:15p.m  Packard 101

In communications, separating source coding from channel coding is a
standard form of modularization. It makes things simple for us, and
furthermore Shannon's Separation Principle allows us to feel that we
are not going to suffer for it. For practical purposes, separation
leads to big toolboxes of reusable tools.

I'll examine certain communication scenarios and draw conclusion on
the existing toolboxes. I'll show that there is life beyond
multiresolution.  In particular, in some communication scenarios, the
information available at the source decoder is a subset of a small
number of chunks of data. Then, the right tools for source coding are
not the conventional ones, but rather multiple description codes. I'll
conclude with demonstrations of multiple description speech and audio
coders.

---------------------------

Jelena Kovacevic received her PhD from Columbia University with Martin
Vetterli in 1991 and has been at Bell Labs (now Lucent Technologies)
since as well as an Adjunct Professor at Columbia. She is coauthor with
Vetterli of "Wavelets and Subband Coding," Prentice Hall,
Editor-in-Chief of the IEEE Transactions on Image Processing, and an
IEEE Fellow.




More information about the labmembers mailing list