Super-node Normalized Belief Propagation For Probabilistic Systems
- Detailed Technology Description
- None
- Supplementary Information
- Patent Number: US6745157B1
Application Number: US2000586281A
Inventor: Weiss, Yair | Freeman, William T. | Yedidia, Jonathan S.
Priority Date: 2 Jun 2000
Priority Number: US6745157B1
Application Date: 2 Jun 2000
Publication Date: 1 Jun 2004
IPC Current: G06F001718 | G06F001710 | H04L001224
US Class: 703002 | 703013 | 706014 | 706015 | 706022 | 706045
Assignee Applicant: Mitsubishi Electric Research Laboratories Incmbridge
Title: Super-node normalized belief propagation for probabilistic systems
Usefulness: Super-node normalized belief propagation for probabilistic systems
Summary: For approximation of both marginal probabilities and MAP states in Markov networks.
Novelty: Probability system modeling for Markov networks uses message passing within loops and finds states at each node which maximizes approximation over the entire network
- Industry
- ICT/Telecom
- Sub Category
- Telecommunication
- Application No.
- 6745157
- Others
-
Tech ID/UC Case
16971/2000-097-0
Related Cases
2000-097-0
- *Abstract
-
Researchers at the University of California, Berkeley have developed an improved method of modeling probabilistic systems, in particular such systems using belief progagation in a Markov network.
- *IP Issue Date
- Jun 1, 2004
- *Principal Investigator
-
Name: WILLIAM FREEMAN
Department:
Name: YAIR WEISS
Department:
Name: JONATHAN YEDIDIA
Department:
- Country/Region
- USA
For more information, please click Here
