Hierarchical variational models

Rajesh Ranganath, Dustin Tran, David M. Blei

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

Abstract

Black box variational inference allows researchers to easily prototype and evaluate an array of models. Recent advances allow such algorithms to scale to high dimensions. However, a central question remains: How to specify an expressive variational distribution that maintains efficient computation? To address this, we develop hierarchical variational models (HVMS). HVMS augment a variational approximation with a prior on its parameters, which allows it to capture complex structure for both discrete and continuous latent variables. The algorithm we develop is black box, can be used for any HVM, and has the same computational efficiency as the original approximation. We study hvms on a variety of deep discrete latent variable models. HVMs generalize other expressive variational distributions and maintains higher fidelity to the posterior.

Original languageEnglish (US)
Title of host publication33rd International Conference on Machine Learning, ICML 2016
PublisherInternational Machine Learning Society (IMLS)
Pages515-528
Number of pages14
Volume1
ISBN (Electronic)9781510829008
StatePublished - Jan 1 2016
Event33rd International Conference on Machine Learning, ICML 2016 - New York City, United States
Duration: Jun 19 2016Jun 24 2016

Other

Other33rd International Conference on Machine Learning, ICML 2016
CountryUnited States
CityNew York City
Period6/19/166/24/16

Fingerprint

Computational efficiency

ASJC Scopus subject areas

  • Artificial Intelligence
  • Software
  • Computer Networks and Communications

Cite this

Ranganath, R., Tran, D., & Blei, D. M. (2016). Hierarchical variational models. In 33rd International Conference on Machine Learning, ICML 2016 (Vol. 1, pp. 515-528). International Machine Learning Society (IMLS).

Hierarchical variational models. / Ranganath, Rajesh; Tran, Dustin; Blei, David M.

33rd International Conference on Machine Learning, ICML 2016. Vol. 1 International Machine Learning Society (IMLS), 2016. p. 515-528.

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

Ranganath, R, Tran, D & Blei, DM 2016, Hierarchical variational models. in 33rd International Conference on Machine Learning, ICML 2016. vol. 1, International Machine Learning Society (IMLS), pp. 515-528, 33rd International Conference on Machine Learning, ICML 2016, New York City, United States, 6/19/16.
Ranganath R, Tran D, Blei DM. Hierarchical variational models. In 33rd International Conference on Machine Learning, ICML 2016. Vol. 1. International Machine Learning Society (IMLS). 2016. p. 515-528
Ranganath, Rajesh ; Tran, Dustin ; Blei, David M. / Hierarchical variational models. 33rd International Conference on Machine Learning, ICML 2016. Vol. 1 International Machine Learning Society (IMLS), 2016. pp. 515-528
@inproceedings{13bd002764df4aa2929229c8983d5119,
title = "Hierarchical variational models",
abstract = "Black box variational inference allows researchers to easily prototype and evaluate an array of models. Recent advances allow such algorithms to scale to high dimensions. However, a central question remains: How to specify an expressive variational distribution that maintains efficient computation? To address this, we develop hierarchical variational models (HVMS). HVMS augment a variational approximation with a prior on its parameters, which allows it to capture complex structure for both discrete and continuous latent variables. The algorithm we develop is black box, can be used for any HVM, and has the same computational efficiency as the original approximation. We study hvms on a variety of deep discrete latent variable models. HVMs generalize other expressive variational distributions and maintains higher fidelity to the posterior.",
author = "Rajesh Ranganath and Dustin Tran and Blei, {David M.}",
year = "2016",
month = "1",
day = "1",
language = "English (US)",
volume = "1",
pages = "515--528",
booktitle = "33rd International Conference on Machine Learning, ICML 2016",
publisher = "International Machine Learning Society (IMLS)",

}

TY - GEN

T1 - Hierarchical variational models

AU - Ranganath, Rajesh

AU - Tran, Dustin

AU - Blei, David M.

PY - 2016/1/1

Y1 - 2016/1/1

N2 - Black box variational inference allows researchers to easily prototype and evaluate an array of models. Recent advances allow such algorithms to scale to high dimensions. However, a central question remains: How to specify an expressive variational distribution that maintains efficient computation? To address this, we develop hierarchical variational models (HVMS). HVMS augment a variational approximation with a prior on its parameters, which allows it to capture complex structure for both discrete and continuous latent variables. The algorithm we develop is black box, can be used for any HVM, and has the same computational efficiency as the original approximation. We study hvms on a variety of deep discrete latent variable models. HVMs generalize other expressive variational distributions and maintains higher fidelity to the posterior.

AB - Black box variational inference allows researchers to easily prototype and evaluate an array of models. Recent advances allow such algorithms to scale to high dimensions. However, a central question remains: How to specify an expressive variational distribution that maintains efficient computation? To address this, we develop hierarchical variational models (HVMS). HVMS augment a variational approximation with a prior on its parameters, which allows it to capture complex structure for both discrete and continuous latent variables. The algorithm we develop is black box, can be used for any HVM, and has the same computational efficiency as the original approximation. We study hvms on a variety of deep discrete latent variable models. HVMs generalize other expressive variational distributions and maintains higher fidelity to the posterior.

UR - http://www.scopus.com/inward/record.url?scp=84997769529&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84997769529&partnerID=8YFLogxK

M3 - Conference contribution

VL - 1

SP - 515

EP - 528

BT - 33rd International Conference on Machine Learning, ICML 2016

PB - International Machine Learning Society (IMLS)

ER -