|
 TITLE: Statistical Inference and Learning with Loops ... and Physics
SPEAKER: Misha Chertkov (Los Alamos National Laboratory)
DATE: 3:00 - 4:00 PM, Monday, December 1, 2008
LOCATION: Sigma, 1L
ABSTRACT:
There has been an explosion of interest in the past decade to statistical
problems related to computer science and information processing, such as new
decoding paradigms for high-volume communication and storage, problems in
search and counting through a huge set of combinatorial constraints,
etc. Novel ideas in analysis of complexity and development of approximate but
systematically improvable algorithms are required for the hard statistical
problems. Since the central task of statistical physics is to describe how
complex behavior emerges from the interaction of a large number of basic
elements, its tools and concepts are proven valuable in these emerging
disciplines.
In this talk I will discuss one such tool, coined Loop Calculus, brought from
physics into the field of statistical inference and learning. Loop Calculus
allows one to express the solution of a general statistical inference problem
exactly via a solution of the so-called Belief Propagation equations. I will
explain its main concept, its algorithmic utility (with examples from error
correction and particle tracking), and future challenges of the Loop Calculus
approach.
BIOGRAPHY:
Dr. Chertkov's areas of interest include statistical and mathematical physics
applied to information theory, computer science, hydrodynamics, optics and
bio-physics. Dr. Chertkov received his Ph.D. in physics from the Weizmann
Institute of Science in 1996, and his M.Sc. in physics from Novosibirsk State
University in 1990. After his Ph.D., Dr. Chertkov spent three years at
Princeton University as a R.H. Dicke Fellow in the Department of Physics. He
joined Los Alamos National Lab in 1999, initially as a J.R. Oppenheimer Fellow
in the Theoretical Division. He is now TSM-4 in the same division.
Dr. Chertkov has published 80 papers in these research areas. He leads
"Physics of Algorithms" Directed Research project at LANL.
|
|
|
printable version
