NIPS workshop on Propagation Algorithms on Graphs with Cycles

NIPS*2002 post conference workshop on

Propagation Algorithms on Graphs with Cycles:
Theory and Applications

December 13th and 14th, 2002, Whistler

Program

December 13th, 2002

Morning Session (7:30 - 10:30)

7:30	S. Ikeda, Opening Remarks
7:35-8:00	I. Rish, IBM T.J. Watson Research Center
7:35-8:00	From constraint propagation to belief propagation: local inference with nearly- deterministic probabilities
8:00-8:25	L. Ortiz, Univ. of Pennsylvania
8:00-8:25	Nash propagation for graphical games and its relation to belief propagation and CSPs
-- Coffee break --
8:40-9:05	T. Tanaka, Tokyo Metropolitan Univ.
8:40-9:05	Approximate belief propagation, density evolution, and statistical neurodynamics for multiuser detection of code- division multiple- access systems
9:05-9:30	Y. Kabashima, Tokyo Inst. of Tech.
9:05-9:30	A mean field approach to the CDMA multiuser detection
9:30-10:30	Poster viewing

Evening Session (16:00-19:00)

16:00-16:25	E. Sudderth, Massachusetts Inst. of Tech.
	Nonparametric belief propagation
16:25-16:50	S. Ikeda, Kyushu Inst. of Tech.
	Loopy belief propagation from information geometrical viewpoint
-- Coffee break --
17:05-17:30	S. Tatikonda, Yale Univ.
	Convergence conditions for the loopy belief propagation algorithm
17:30-17:55	M. Wainwright, Univ. of Calfornia at Berkeley
	Convex and semidefinite relaxations for approximate inference in graphs with cycles
17:55-18:10	S. Thrun, Carnegie Mellon Univ.
	Learning sparse Gaussian Markov random field in robotics
18:10-19:00	Poster viewing

December 14th, 2002

Morning Session (7:30 - 10:30)

7:30-7:55	H. Kappen, Univ. of Nijmegen
	The cluster variation method for approximate inference
7:55-8:20	K. Tanaka, Tohoku Univ.
	Cluster variation method for correlation function of probabilistic model with loopy graphical structure
-- Coffee break --
8:35-9:00	M. Welling, Univ. of Toronto
	Propagation rules for linear response estimates of joint pairwise probabilities
9:00-9:25	J. Yedidia, Mitsubishi Electric Research Laboratories
	Generating code representations suitable for belief propagation decoding
9:25-9:40	O. Winther, Technical Univ. of Denmark
	Towards more accurate free energies
9:25-10:30	Poster viewing

Evening Session (16:00-19:00)

16:00-16:25	T. Heskes, Univ. of Nijmegen
16:00-16:25	Loopy belief propagation for approximate inference in hybrid Bayesian networks
16:25-16:50	R. Palanki, California Inst. of Tech.
16:25-16:50	A provably convergent belief optimization algorithm
-- Coffee break --
17:05-17:30	A. Yuille, Univ. of California, Los Angels
17:05-17:30	CCCP, MIME, and the max clique problem
17:30-17:55	W. Freeman, Massachusetts Inst. of Tech.
17:30-17:55	Applications of belief propagation in vision and graphics
17:55-18:55	Poster viewing
18:55	M. Welling, Closing Remarks

Last modified: Thu Dec 5 18:13:34 JST 2002