Broad Area Colloquium For AI-Geometry-Graphics-Robotics-Vision
Microsoft Research: Bridging the Analog and Digital Worlds
Microsoft Research Redmond
Monday, May 10, 2004, 4:15PM
Increasingly computers are being used in applications where voice,
audio, and scanned images are important data types. In this talk I will
describe a series of projects at Microsoft Research that address gaining
structured information from these real world (analog) signal sources.
Speech recognition is a classic signal interpretation problem, and we
will describe recent research in microphones and sensors for robust
recognition. Next, we will discuss audio fingerprinting, a technique
for recognizing songs in online streams or on a PC that can be used in a
variety of personal entertainment scenarios. Finally, we will discuss
issues in document interpretation: extraction of structural, semantic
information from scanned documents such as faxes or printed documents.
All these projects require a combination of signal processing/machine
learning techniques and large datasets; we will discuss the associated
methodological challenges and practical applications.
About the Speaker
Jack Breese is a Director of Microsoft Research in Redmond, WA, where he
oversees work on intelligent systems including data management, machine
learning, communications, collaboration, natural language processing,
communities, document understanding, and adaptive systems. Previously
he was a founding member of the Decision Theory and Adaptive Systems
research group at Microsoft Research, where he developed basic
technologies and tools for user modeling, intelligent diagnostics,
adaptive systems, recommender systems, e-ecommerce, and datamining.
Breese received a doctorate from Stanford University in 1987 in
Engineering-Economic Systems (now Management Science and Engineering)
and joined Microsoft Research in 1993.
Back to the Colloquium Page