Node Relevance Determination

Neil D. Lawrence
The {XI}th Italian Workshop on Neural Networks, Springer-Verlag , 2001.

Abstract

Hierarchical Bayesian inference in parameterised models offers an approach for controlling complexity. In this paper we utilise a novel prior for the leaning of a model’s structure. We call the prior *node relevance determination*. It is applicable in a range of models including sigmoid belief networks and Boltzmann machines. We demonstrate how the approach may be applied to determine structure in a multi-layer perceptron.

Cite this Paper

BibTeX
@InProceedings{Lawrence:nrd01, title = {Node Relevance Determination}, author = {Lawrence, Neil D.}, booktitle = {The {XI}th Italian Workshop on Neural Networks}, year = {2001}, editor = {Marinaro, Maria and Tagliaferri, Roberto}, publisher = {Springer-Verlag}, pdf = {http://www.thelawrences.net/neil/structure.pdf}, url = {http://inverseprobability.com/publications/lawrence-nrd01.html}, abstract = {Hierarchical Bayesian inference in parameterised models offers an approach for controlling complexity. In this paper we utilise a novel prior for the leaning of a model’s structure. We call the prior *node relevance determination*. It is applicable in a range of models including sigmoid belief networks and Boltzmann machines. We demonstrate how the approach may be applied to determine structure in a multi-layer perceptron.} }
Endnote
%0 Conference Paper %T Node Relevance Determination %A Neil D. Lawrence %B The {XI}th Italian Workshop on Neural Networks %D 2001 %E Maria Marinaro %E Roberto Tagliaferri %F Lawrence:nrd01 %I Springer-Verlag %U http://inverseprobability.com/publications/lawrence-nrd01.html %X Hierarchical Bayesian inference in parameterised models offers an approach for controlling complexity. In this paper we utilise a novel prior for the leaning of a model’s structure. We call the prior *node relevance determination*. It is applicable in a range of models including sigmoid belief networks and Boltzmann machines. We demonstrate how the approach may be applied to determine structure in a multi-layer perceptron.
RIS
TY - CPAPER TI - Node Relevance Determination AU - Neil D. Lawrence BT - The {XI}th Italian Workshop on Neural Networks DA - 2001/01/01 ED - Maria Marinaro ED - Roberto Tagliaferri ID - Lawrence:nrd01 PB - Springer-Verlag L1 - http://www.thelawrences.net/neil/structure.pdf UR - http://inverseprobability.com/publications/lawrence-nrd01.html AB - Hierarchical Bayesian inference in parameterised models offers an approach for controlling complexity. In this paper we utilise a novel prior for the leaning of a model’s structure. We call the prior *node relevance determination*. It is applicable in a range of models including sigmoid belief networks and Boltzmann machines. We demonstrate how the approach may be applied to determine structure in a multi-layer perceptron. ER -
APA
Lawrence, N.D.. (2001). Node Relevance Determination. The {XI}th Italian Workshop on Neural Networks Available from http://inverseprobability.com/publications/lawrence-nrd01.html.

Related Material