Advanced mean field methods: theory and practice

Manfred Opper (Editor), David Saad (Editor)

Research output: Book/ReportBook

Abstract

A major problem in modern probabilistic modeling is the huge computational complexity involved in typical calculations with multivariate probability distributions when the number of random variables is large. Because exact computations are infeasible in such cases and Monte Carlo sampling techniques may reach their limits, there is a need for methods that allow for efficient approximate computations. One of the simplest approximations is based on the mean field method, which has a long history in statistical physics. The method is widely used, particularly in the growing field of graphical models. Researchers from disciplines such as statistical physics, computer science, and mathematical statistics are studying ways to improve this and related methods and are exploring novel application areas. Leading approaches include the variational approach, which goes beyond factorizable distributions to achieve systematic improvements; the TAP (Thouless-Anderson-Palmer) approach, which incorporates correlations by including effective reaction terms in the mean field theory; and the more general methods of graphical models. Bringing together ideas and techniques from these diverse disciplines, this book covers the theoretical foundations of advanced mean field methods, explores the relation between the different approaches, examines the quality of the approximation obtained, and demonstrates their application to various areas of probabilistic modeling.
Original languageEnglish
Place of PublicationCambridge, Massachusetts (US)
PublisherMIT
Number of pages287
ISBN (Print)0262150549
Publication statusPublished - Feb 2001

Publication series

NameNeural Information Processing
PublisherMassachusetts Institute of Technology Press (MIT Press)

Fingerprint

field method
physics
modeling
method
history
distribution

Keywords

  • probabilistic modeling
  • multivariate probability distributions
  • efficient approximate computations
  • TAP approach

Cite this

Opper, M., & Saad, D. (Eds.) (2001). Advanced mean field methods: theory and practice. (Neural Information Processing). Cambridge, Massachusetts (US): MIT.
Opper, Manfred (Editor) ; Saad, David (Editor). / Advanced mean field methods : theory and practice. Cambridge, Massachusetts (US) : MIT, 2001. 287 p. (Neural Information Processing).
@book{32abdd69aa304559b1b5db4bf73ae158,
title = "Advanced mean field methods: theory and practice",
abstract = "A major problem in modern probabilistic modeling is the huge computational complexity involved in typical calculations with multivariate probability distributions when the number of random variables is large. Because exact computations are infeasible in such cases and Monte Carlo sampling techniques may reach their limits, there is a need for methods that allow for efficient approximate computations. One of the simplest approximations is based on the mean field method, which has a long history in statistical physics. The method is widely used, particularly in the growing field of graphical models. Researchers from disciplines such as statistical physics, computer science, and mathematical statistics are studying ways to improve this and related methods and are exploring novel application areas. Leading approaches include the variational approach, which goes beyond factorizable distributions to achieve systematic improvements; the TAP (Thouless-Anderson-Palmer) approach, which incorporates correlations by including effective reaction terms in the mean field theory; and the more general methods of graphical models. Bringing together ideas and techniques from these diverse disciplines, this book covers the theoretical foundations of advanced mean field methods, explores the relation between the different approaches, examines the quality of the approximation obtained, and demonstrates their application to various areas of probabilistic modeling.",
keywords = "probabilistic modeling, multivariate probability distributions, efficient approximate computations, TAP approach",
editor = "Manfred Opper and David Saad",
year = "2001",
month = "2",
language = "English",
isbn = "0262150549",
series = "Neural Information Processing",
publisher = "MIT",

}

Opper, M & Saad, D (eds) 2001, Advanced mean field methods: theory and practice. Neural Information Processing, MIT, Cambridge, Massachusetts (US).

Advanced mean field methods : theory and practice. / Opper, Manfred (Editor); Saad, David (Editor).

Cambridge, Massachusetts (US) : MIT, 2001. 287 p. (Neural Information Processing).

Research output: Book/ReportBook

TY - BOOK

T1 - Advanced mean field methods

T2 - theory and practice

A2 - Opper, Manfred

A2 - Saad, David

PY - 2001/2

Y1 - 2001/2

N2 - A major problem in modern probabilistic modeling is the huge computational complexity involved in typical calculations with multivariate probability distributions when the number of random variables is large. Because exact computations are infeasible in such cases and Monte Carlo sampling techniques may reach their limits, there is a need for methods that allow for efficient approximate computations. One of the simplest approximations is based on the mean field method, which has a long history in statistical physics. The method is widely used, particularly in the growing field of graphical models. Researchers from disciplines such as statistical physics, computer science, and mathematical statistics are studying ways to improve this and related methods and are exploring novel application areas. Leading approaches include the variational approach, which goes beyond factorizable distributions to achieve systematic improvements; the TAP (Thouless-Anderson-Palmer) approach, which incorporates correlations by including effective reaction terms in the mean field theory; and the more general methods of graphical models. Bringing together ideas and techniques from these diverse disciplines, this book covers the theoretical foundations of advanced mean field methods, explores the relation between the different approaches, examines the quality of the approximation obtained, and demonstrates their application to various areas of probabilistic modeling.

AB - A major problem in modern probabilistic modeling is the huge computational complexity involved in typical calculations with multivariate probability distributions when the number of random variables is large. Because exact computations are infeasible in such cases and Monte Carlo sampling techniques may reach their limits, there is a need for methods that allow for efficient approximate computations. One of the simplest approximations is based on the mean field method, which has a long history in statistical physics. The method is widely used, particularly in the growing field of graphical models. Researchers from disciplines such as statistical physics, computer science, and mathematical statistics are studying ways to improve this and related methods and are exploring novel application areas. Leading approaches include the variational approach, which goes beyond factorizable distributions to achieve systematic improvements; the TAP (Thouless-Anderson-Palmer) approach, which incorporates correlations by including effective reaction terms in the mean field theory; and the more general methods of graphical models. Bringing together ideas and techniques from these diverse disciplines, this book covers the theoretical foundations of advanced mean field methods, explores the relation between the different approaches, examines the quality of the approximation obtained, and demonstrates their application to various areas of probabilistic modeling.

KW - probabilistic modeling

KW - multivariate probability distributions

KW - efficient approximate computations

KW - TAP approach

UR - http://mitpress.mit.edu/catalog/item/default.asp?ttype=2&tid=3847

M3 - Book

SN - 0262150549

T3 - Neural Information Processing

BT - Advanced mean field methods

PB - MIT

CY - Cambridge, Massachusetts (US)

ER -

Opper M, (ed.), Saad D, (ed.). Advanced mean field methods: theory and practice. Cambridge, Massachusetts (US): MIT, 2001. 287 p. (Neural Information Processing).