Mixture Representations for Inference and Learning in Boltzmann Machines

Neil D. Lawrence, Christopher M. Bishop, Michael I. Jordan
Uncertainty in Artificial Intelligence, Morgan Kauffman 14:320-327, 1998.

Abstract

Boltzmann machines are undirected graphical models with two-state stochastic variables, in which the logarithms of the clique potentials are quadratic functions of the node states. They have been widely studied in the neural computing literature, although their practical applicability has been limited by the difficulty of finding an effective learning algorithm. One well-established approach, known as mean field theory, represents the stochastic distribution using a factorized approximation. However, the corresponding learning algorithm often fails to find a good solution. We conjecture that this is due to the implicit uni-modality of the mean field approximation which is therefore unable to capture multi-modality in the true distribution. In this paper we use variational methods to approximate the stochastic distribution using multi-modal *mixtures* of factorized distributions. We present results for both inference and learning to demonstrate the effectiveness of this approach.

Cite this Paper

BibTeX
@InProceedings{Lawrence:mixtures98, title = {Mixture Representations for Inference and Learning in {B}oltzmann Machines}, author = {Lawrence, Neil D. and Bishop, Christopher M. and Jordan, Michael I.}, booktitle = {Uncertainty in Artificial Intelligence}, pages = {320--327}, year = {1998}, editor = {Cooper, Gregory F. and Moral, Serafín}, volume = {14}, address = {San Francisco, CA}, publisher = {Morgan Kauffman}, pdf = {https://inverseprobability.com/publications/files/boltzmann.pdf}, url = {http://inverseprobability.com/publications/lawrence-mixtures98.html}, abstract = {Boltzmann machines are undirected graphical models with two-state stochastic variables, in which the logarithms of the clique potentials are quadratic functions of the node states. They have been widely studied in the neural computing literature, although their practical applicability has been limited by the difficulty of finding an effective learning algorithm. One well-established approach, known as mean field theory, represents the stochastic distribution using a factorized approximation. However, the corresponding learning algorithm often fails to find a good solution. We conjecture that this is due to the implicit uni-modality of the mean field approximation which is therefore unable to capture multi-modality in the true distribution. In this paper we use variational methods to approximate the stochastic distribution using multi-modal *mixtures* of factorized distributions. We present results for both inference and learning to demonstrate the effectiveness of this approach. } }
Endnote
%0 Conference Paper %T Mixture Representations for Inference and Learning in Boltzmann Machines %A Neil D. Lawrence %A Christopher M. Bishop %A Michael I. Jordan %B Uncertainty in Artificial Intelligence %D 1998 %E Gregory F. Cooper %E Serafín Moral %F Lawrence:mixtures98 %I Morgan Kauffman %P 320--327 %U http://inverseprobability.com/publications/lawrence-mixtures98.html %V 14 %X Boltzmann machines are undirected graphical models with two-state stochastic variables, in which the logarithms of the clique potentials are quadratic functions of the node states. They have been widely studied in the neural computing literature, although their practical applicability has been limited by the difficulty of finding an effective learning algorithm. One well-established approach, known as mean field theory, represents the stochastic distribution using a factorized approximation. However, the corresponding learning algorithm often fails to find a good solution. We conjecture that this is due to the implicit uni-modality of the mean field approximation which is therefore unable to capture multi-modality in the true distribution. In this paper we use variational methods to approximate the stochastic distribution using multi-modal *mixtures* of factorized distributions. We present results for both inference and learning to demonstrate the effectiveness of this approach.
RIS
TY - CPAPER TI - Mixture Representations for Inference and Learning in Boltzmann Machines AU - Neil D. Lawrence AU - Christopher M. Bishop AU - Michael I. Jordan BT - Uncertainty in Artificial Intelligence DA - 1998/01/01 ED - Gregory F. Cooper ED - Serafín Moral ID - Lawrence:mixtures98 PB - Morgan Kauffman VL - 14 SP - 320 EP - 327 L1 - https://inverseprobability.com/publications/files/boltzmann.pdf UR - http://inverseprobability.com/publications/lawrence-mixtures98.html AB - Boltzmann machines are undirected graphical models with two-state stochastic variables, in which the logarithms of the clique potentials are quadratic functions of the node states. They have been widely studied in the neural computing literature, although their practical applicability has been limited by the difficulty of finding an effective learning algorithm. One well-established approach, known as mean field theory, represents the stochastic distribution using a factorized approximation. However, the corresponding learning algorithm often fails to find a good solution. We conjecture that this is due to the implicit uni-modality of the mean field approximation which is therefore unable to capture multi-modality in the true distribution. In this paper we use variational methods to approximate the stochastic distribution using multi-modal *mixtures* of factorized distributions. We present results for both inference and learning to demonstrate the effectiveness of this approach. ER -
APA
Lawrence, N.D., Bishop, C.M. & Jordan, M.I.. (1998). Mixture Representations for Inference and Learning in Boltzmann Machines. Uncertainty in Artificial Intelligence 14:320-327 Available from http://inverseprobability.com/publications/lawrence-mixtures98.html.

Related Material