A Bayesian Matrix Factorization Model for Relational Data

Ajit Singh, Geoffrey Gordon
Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence, PMLR R8:555-562, 2010.

Abstract

Relational learning can be used to augment one data source with other correlated sources of information, to improve predictive accuracy. We frame a large class of relational learning problems as matrix factorization problems, and propose a hierarchical Bayesian model. Training our Bayesian model using random-walk Metropolis-Hastings is impractically slow, and so we develop a block Metropolis- Hastings sampler which uses the gradient and Hessian of the likelihood to dynamically tune the proposal. We demonstrate that a predictive model of brain response to stimuli can be improved by augmenting it with side information about the stimuli.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR8-singh10a, title = {A {B}ayesian Matrix Factorization Model for Relational Data}, author = {Singh, Ajit and Gordon, Geoffrey}, booktitle = {Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence}, pages = {555--562}, year = {2010}, editor = {Grünwald, Peter and Spirtes, Peter}, volume = {R8}, series = {Proceedings of Machine Learning Research}, month = {08--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r8/main/assets/singh10a/singh10a.pdf}, url = {https://proceedings.mlr.press/r8/singh10a.html}, abstract = {Relational learning can be used to augment one data source with other correlated sources of information, to improve predictive accuracy. We frame a large class of relational learning problems as matrix factorization problems, and propose a hierarchical Bayesian model. Training our Bayesian model using random-walk Metropolis-Hastings is impractically slow, and so we develop a block Metropolis- Hastings sampler which uses the gradient and Hessian of the likelihood to dynamically tune the proposal. We demonstrate that a predictive model of brain response to stimuli can be improved by augmenting it with side information about the stimuli.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T A Bayesian Matrix Factorization Model for Relational Data %A Ajit Singh %A Geoffrey Gordon %B Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2010 %E Peter Grünwald %E Peter Spirtes %F pmlr-vR8-singh10a %I PMLR %P 555--562 %U https://proceedings.mlr.press/r8/singh10a.html %V R8 %X Relational learning can be used to augment one data source with other correlated sources of information, to improve predictive accuracy. We frame a large class of relational learning problems as matrix factorization problems, and propose a hierarchical Bayesian model. Training our Bayesian model using random-walk Metropolis-Hastings is impractically slow, and so we develop a block Metropolis- Hastings sampler which uses the gradient and Hessian of the likelihood to dynamically tune the proposal. We demonstrate that a predictive model of brain response to stimuli can be improved by augmenting it with side information about the stimuli. %Z Reissued by PMLR on 04 October 2026.
APA
Singh, A. & Gordon, G.. (2010). A Bayesian Matrix Factorization Model for Relational Data. Proceedings of the 26th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R8:555-562 Available from https://proceedings.mlr.press/r8/singh10a.html. Reissued by PMLR on 04 October 2026.

Related Material