Majorization for CRFs and latent likelihoods (Supplementary material)

Tony Jebara, Anna Choromanska

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This supplement presents additional details in support of the full article. These include the application of the majorization method to maximum entropy problems. It also contains proofs of the various theorems, in particular, a guarantee that the bound majorizes the partition function. In addition, a proof is provided guaranteeing convergence on (non-latent) maximum conditional likelihood problems. The supplement also contains supporting lemmas that show the bound remains applicable in constrained optimization problems. The supplement then proves the soundness of the junction tree implementation of the bound for graphical mod-els with large n. It also proves the soundness of the low-rank implementation of the bound for problems with large d. Finally, the supplement contains additional experiments and figures to provide further empirical support for the majorization methodology.

Original languageEnglish (US)
Title of host publicationAdvances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012, NIPS 2012
Volume1
StatePublished - 2012
Event26th Annual Conference on Neural Information Processing Systems 2012, NIPS 2012 - Lake Tahoe, NV, United States
Duration: Dec 3 2012Dec 6 2012

Other

Other26th Annual Conference on Neural Information Processing Systems 2012, NIPS 2012
CountryUnited States
CityLake Tahoe, NV
Period12/3/1212/6/12

Fingerprint

Constrained optimization
Maximum likelihood
Entropy
Experiments

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Information Systems
  • Signal Processing

Cite this

Jebara, T., & Choromanska, A. (2012). Majorization for CRFs and latent likelihoods (Supplementary material). In Advances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012, NIPS 2012 (Vol. 1)

Majorization for CRFs and latent likelihoods (Supplementary material). / Jebara, Tony; Choromanska, Anna.

Advances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012, NIPS 2012. Vol. 1 2012.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Jebara, T & Choromanska, A 2012, Majorization for CRFs and latent likelihoods (Supplementary material). in Advances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012, NIPS 2012. vol. 1, 26th Annual Conference on Neural Information Processing Systems 2012, NIPS 2012, Lake Tahoe, NV, United States, 12/3/12.
Jebara T, Choromanska A. Majorization for CRFs and latent likelihoods (Supplementary material). In Advances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012, NIPS 2012. Vol. 1. 2012
Jebara, Tony ; Choromanska, Anna. / Majorization for CRFs and latent likelihoods (Supplementary material). Advances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012, NIPS 2012. Vol. 1 2012.
@inproceedings{695d896d791c43268c5905305823a48e,
title = "Majorization for CRFs and latent likelihoods (Supplementary material)",
abstract = "This supplement presents additional details in support of the full article. These include the application of the majorization method to maximum entropy problems. It also contains proofs of the various theorems, in particular, a guarantee that the bound majorizes the partition function. In addition, a proof is provided guaranteeing convergence on (non-latent) maximum conditional likelihood problems. The supplement also contains supporting lemmas that show the bound remains applicable in constrained optimization problems. The supplement then proves the soundness of the junction tree implementation of the bound for graphical mod-els with large n. It also proves the soundness of the low-rank implementation of the bound for problems with large d. Finally, the supplement contains additional experiments and figures to provide further empirical support for the majorization methodology.",
author = "Tony Jebara and Anna Choromanska",
year = "2012",
language = "English (US)",
isbn = "9781627480031",
volume = "1",
booktitle = "Advances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012, NIPS 2012",

}

TY - GEN

T1 - Majorization for CRFs and latent likelihoods (Supplementary material)

AU - Jebara, Tony

AU - Choromanska, Anna

PY - 2012

Y1 - 2012

N2 - This supplement presents additional details in support of the full article. These include the application of the majorization method to maximum entropy problems. It also contains proofs of the various theorems, in particular, a guarantee that the bound majorizes the partition function. In addition, a proof is provided guaranteeing convergence on (non-latent) maximum conditional likelihood problems. The supplement also contains supporting lemmas that show the bound remains applicable in constrained optimization problems. The supplement then proves the soundness of the junction tree implementation of the bound for graphical mod-els with large n. It also proves the soundness of the low-rank implementation of the bound for problems with large d. Finally, the supplement contains additional experiments and figures to provide further empirical support for the majorization methodology.

AB - This supplement presents additional details in support of the full article. These include the application of the majorization method to maximum entropy problems. It also contains proofs of the various theorems, in particular, a guarantee that the bound majorizes the partition function. In addition, a proof is provided guaranteeing convergence on (non-latent) maximum conditional likelihood problems. The supplement also contains supporting lemmas that show the bound remains applicable in constrained optimization problems. The supplement then proves the soundness of the junction tree implementation of the bound for graphical mod-els with large n. It also proves the soundness of the low-rank implementation of the bound for problems with large d. Finally, the supplement contains additional experiments and figures to provide further empirical support for the majorization methodology.

UR - http://www.scopus.com/inward/record.url?scp=84877784438&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84877784438&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9781627480031

VL - 1

BT - Advances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012, NIPS 2012

ER -