While mathematical in nature, information theory addresses at least
three basic engineering problems: data compression, coding for error
correction, and certain aspects of cryptography. It is therefore not
surprising that an industrial research lab would want to maintain core
expertise in technologies that were crucial in enabling the multimedia
revolution. But if the target is the practical applications - the
algorithms, the industry standards - why research the mathematical
foundations - the mathematical models, the fundamental bounds? In this
talk, I will try to address this question by presenting "a few years"
in the life of the Information Theory Research group at HP Labs. How
does the group "research the mathematical foundations and practical
applications of information theory, generating intellectual property
and technology for HP through the advancement of scientific knowledge
in these areas," as its "mission" states? How does its approach differ
from the prevalent one among information theorists in the academia?
From theorems in universal coding and universal denoising to industrial
standards on image compression and patents on image denoising, from
capacity calculations in constrained coding to DVD standards, we will
discuss the many faces of information theory.