EM optimization of latent-variable density models

Christopher M. Bishop; M. Svens'en; Christopher K. I. Williams

EM optimization of latent-variable density models

Christopher M. Bishop, M. Svens'en, Christopher K. I. Williams

Research output: Chapter in Book/Published conference output › Chapter

Abstract

There is currently considerable interest in developing general non-linear density models based on latent, or hidden, variables. Such models have the ability to discover the presence of a relatively small number of underlying `causes' which, acting in combination, give rise to the apparent complexity of the observed data set. Unfortunately, to train such models generally requires large computational effort. In this paper we introduce a novel latent variable algorithm which retains the general non-linear capabilities of previous models but which uses a training procedure based on the EM algorithm. We demonstrate the performance of the model on a toy problem and on data from flow diagnostics for a multi-phase oil pipeline.

Original language	English
Title of host publication	Advances in Neural Information Processing Systems 8
Editors	D. S. Touretzky, M. C. Mozer, M. E. Hasselmo
Place of Publication	Cambridge, MA
Publisher	MIT
Pages	465-471
Number of pages	7
Volume	8
ISBN (Print)	0262201070
Publication status	Published - Jun 1996
Event	Advances in Neural Information Processing Systems 1996 - Hong Kong, China Duration: 12 Nov 1996 → 14 Nov 1996

Conference

Conference	Advances in Neural Information Processing Systems 1996
Country/Territory	China
City	Hong Kong
Period	12/11/96 → 14/11/96

Bibliographical note

Copyright of the Massachusetts Institute of Technology Press (MIT Press)

Keywords

NCRG

Access to Document

EM optimization of latent-variable density models
Copyright of the Massachusetts Institute of Technology Press (MIT Press)
Final published version, 703 KB

Cite this

@inbook{e63162bf76804f9ca2d3ff6395c2f851,

title = "EM optimization of latent-variable density models",

abstract = "There is currently considerable interest in developing general non-linear density models based on latent, or hidden, variables. Such models have the ability to discover the presence of a relatively small number of underlying `causes' which, acting in combination, give rise to the apparent complexity of the observed data set. Unfortunately, to train such models generally requires large computational effort. In this paper we introduce a novel latent variable algorithm which retains the general non-linear capabilities of previous models but which uses a training procedure based on the EM algorithm. We demonstrate the performance of the model on a toy problem and on data from flow diagnostics for a multi-phase oil pipeline.",

keywords = "NCRG",

author = "Bishop, {Christopher M.} and M. Svens'en and Williams, {Christopher K. I.}",

note = "Copyright of the Massachusetts Institute of Technology Press (MIT Press); Advances in Neural Information Processing Systems 1996 ; Conference date: 12-11-1996 Through 14-11-1996",

year = "1996",

month = jun,

language = "English",

isbn = "0262201070",

volume = "8",

pages = "465--471",

editor = "Touretzky, {D. S.} and Mozer, {M. C.} and Hasselmo, {M. E.}",

booktitle = "Advances in Neural Information Processing Systems 8",

publisher = "MIT",

}

TY - CHAP

T1 - EM optimization of latent-variable density models

AU - Bishop, Christopher M.

AU - Svens'en, M.

AU - Williams, Christopher K. I.

N1 - Copyright of the Massachusetts Institute of Technology Press (MIT Press)

PY - 1996/6

Y1 - 1996/6

N2 - There is currently considerable interest in developing general non-linear density models based on latent, or hidden, variables. Such models have the ability to discover the presence of a relatively small number of underlying `causes' which, acting in combination, give rise to the apparent complexity of the observed data set. Unfortunately, to train such models generally requires large computational effort. In this paper we introduce a novel latent variable algorithm which retains the general non-linear capabilities of previous models but which uses a training procedure based on the EM algorithm. We demonstrate the performance of the model on a toy problem and on data from flow diagnostics for a multi-phase oil pipeline.

AB - There is currently considerable interest in developing general non-linear density models based on latent, or hidden, variables. Such models have the ability to discover the presence of a relatively small number of underlying `causes' which, acting in combination, give rise to the apparent complexity of the observed data set. Unfortunately, to train such models generally requires large computational effort. In this paper we introduce a novel latent variable algorithm which retains the general non-linear capabilities of previous models but which uses a training procedure based on the EM algorithm. We demonstrate the performance of the model on a toy problem and on data from flow diagnostics for a multi-phase oil pipeline.

KW - NCRG

UR - http://mitpress.mit.edu/catalog/item/default.asp?ttype=2&tid=8421

M3 - Chapter

SN - 0262201070

VL - 8

SP - 465

EP - 471

BT - Advances in Neural Information Processing Systems 8

A2 - Touretzky, D. S.

A2 - Mozer, M. C.

A2 - Hasselmo, M. E.

PB - MIT

CY - Cambridge, MA

T2 - Advances in Neural Information Processing Systems 1996

Y2 - 12 November 1996 through 14 November 1996

ER -

EM optimization of latent-variable density models

Abstract

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this