Fast Learning by Bounding Likelihoods in Sigmoid Type Belief Networks

Fast Learning by Bounding Likelihoods in Sigmoid Type Belief Networks (1995)

Cached

Download Links

[www.ai.mit.edu]
[unlser1.unl.csi.cuny.edu]

Other Repositories/Bibliography

DBLP

by Tommi Jaakkola , Lawrence K. Saul , Michael I. Jordan

Venue:	Advances in Neural Information Processing Systems 8
Citations:	21 - 2 self

Versions

Version History

Metadata Version 3

User correction supplied by SystemCorrections

Datum	Value	Source
TITLE	Fast Learning by Bounding Likelihoods in Sigmoid Type Belief Networks	user correction - Legacy Corrections
AUTHOR NAME	Tommi Jaakkola	SVM HeaderParse 0.1
AUTHOR NAME	Lawrence K. Saul	SVM HeaderParse 0.1
AUTHOR NAME	Michael I. Jordan	SVM HeaderParse 0.1
AUTHOR AFFIL	Department of Brain and Cognitive Sciences; Massachusetts Institute of Technology	SVM HeaderParse 0.2
AUTHOR ADDR	Cambridge, MA 02139	SVM HeaderParse 0.1
ABSTRACT	Sigmoid type belief networks, a class of probabilistic neural networks, provide a natural framework for compactly representing probabilistic information in a variety of unsupervised and supervised learning problems. Often the parameters used in these networks need to be learned from examples. Unfortunately, estimating the parameters via exact probabilistic calculations (i.e, the EM-algorithm) is intractable even for networks with fairly small numbers of hidden units. We propose to avoid the infeasibility of the E step by bounding likelihoods instead of computing them exactly. We introduce extended and complementary representations for these networks and show that the estimation of the network parameters can be made fast (reduced to quadratic optimization) by performing the estimation in either of the alternative domains. The complementary networks can be used for continuous density estimation as well. 1 Introduction The appeal of probabilistic networks for knowledge representation, in...	user correction - Legacy Corrections
YEAR	1995	user correction - Legacy Corrections
VENUE	Advances in Neural Information Processing Systems 8	INFERENCE
VENUE TYPE	CONFERENCE	INFERENCE
PAGES	528--534	INFERENCE
VOLUME	8	INFERENCE
CITATIONS	7 found	ParsCit 1.0