Unsupervised Learning of Noisy-OR Bayesian Networks

Yonatan Halpern, David Sontag
Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence, PMLR R11:422-431, 2013.

Abstract

This paper considers the problem of learning the parameters in Bayesian networks of discrete variables with known structure and hidden variables. Previous approaches in these settings typically use expectation maximization; when the network has high treewidth, the required expectations might be approximated using Monte Carlo or variational methods. We show how to avoid inference altogether during learning by giving a polynomial-time algorithm based on the method-of-moments, building upon recent work on learning discrete-valued mixture models. In particular, we show how to learn the parameters for a family of bipartite noisy-or Bayesian networks. In our experimental results, we demonstrate an application of our algorithm to learning QMR-DT, a large Bayesian network used for medical diagnosis. We show that it is possible to fully learn the parameters of QMR-DT even when only the findings are observed in the training data (ground truth diseases unknown).

Cite this Paper

BibTeX
@InProceedings{pmlr-vR11-halpern13a, title = {Unsupervised Learning of Noisy-{OR} {B}ayesian Networks}, author = {Halpern, Yonatan and Sontag, David}, booktitle = {Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence}, pages = {422--431}, year = {2013}, editor = {Nicholson, Ann and Smyth, Padhraic}, volume = {R11}, series = {Proceedings of Machine Learning Research}, month = {12--14 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r11/main/assets/halpern13a/halpern13a.pdf}, url = {https://proceedings.mlr.press/r11/halpern13a.html}, abstract = {This paper considers the problem of learning the parameters in Bayesian networks of discrete variables with known structure and hidden variables. Previous approaches in these settings typically use expectation maximization; when the network has high treewidth, the required expectations might be approximated using Monte Carlo or variational methods. We show how to avoid inference altogether during learning by giving a polynomial-time algorithm based on the method-of-moments, building upon recent work on learning discrete-valued mixture models. In particular, we show how to learn the parameters for a family of bipartite noisy-or Bayesian networks. In our experimental results, we demonstrate an application of our algorithm to learning QMR-DT, a large Bayesian network used for medical diagnosis. We show that it is possible to fully learn the parameters of QMR-DT even when only the findings are observed in the training data (ground truth diseases unknown).}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Unsupervised Learning of Noisy-OR Bayesian Networks %A Yonatan Halpern %A David Sontag %B Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2013 %E Ann Nicholson %E Padhraic Smyth %F pmlr-vR11-halpern13a %I PMLR %P 422--431 %U https://proceedings.mlr.press/r11/halpern13a.html %V R11 %X This paper considers the problem of learning the parameters in Bayesian networks of discrete variables with known structure and hidden variables. Previous approaches in these settings typically use expectation maximization; when the network has high treewidth, the required expectations might be approximated using Monte Carlo or variational methods. We show how to avoid inference altogether during learning by giving a polynomial-time algorithm based on the method-of-moments, building upon recent work on learning discrete-valued mixture models. In particular, we show how to learn the parameters for a family of bipartite noisy-or Bayesian networks. In our experimental results, we demonstrate an application of our algorithm to learning QMR-DT, a large Bayesian network used for medical diagnosis. We show that it is possible to fully learn the parameters of QMR-DT even when only the findings are observed in the training data (ground truth diseases unknown). %Z Reissued by PMLR on 04 October 2026.
APA
Halpern, Y. & Sontag, D.. (2013). Unsupervised Learning of Noisy-OR Bayesian Networks. Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R11:422-431 Available from https://proceedings.mlr.press/r11/halpern13a.html. Reissued by PMLR on 04 October 2026.

Related Material