On Pruning with the MDL Score

Eunice Yuh-Jie Chen; Arthur Choi; Adnan Darwiche

On Pruning with the MDL Score

Eunice Yuh-Jie Chen, Arthur Choi, Adnan Darwiche ;

Proceedings of the Eighth International Conference on Probabilistic Graphical Models, PMLR 52:98-109, 2016.

Abstract

The space of Bayesian network structures is forbiddingly large and hence numerous techniques have been developed to prune this search space, but without eliminating the optimal structure. Such techniques are critical for structure learning to scale to larger datasets with more variables. Prior works exploited properties of the MDL score to prune away large regions of the search space that can be safely ignored by optimal structure learning algorithms. In this paper, we propose new techniques for pruning regions of the search space that can be safely ignored by algorithms that enumerate the k-best Bayesian network structures. Empirically, we show that these techniques allow a state-of-the-art structure enumeration algorithm to scale to datasets with significantly more variables.

Related Material

Download PDF

@InProceedings{pmlr-v52-chen16, title = {On Pruning with the MDL Score}, author = {Eunice Yuh-Jie Chen and Arthur Choi and Adnan Darwiche}, booktitle = {Proceedings of the Eighth International Conference on Probabilistic Graphical Models}, pages = {98--109}, year = {2016}, editor = {Alessandro Antonucci and Giorgio Corani and Cassio Polpo Campos}}, volume = {52}, series = {Proceedings of Machine Learning Research}, address = {Lugano, Switzerland}, month = {06--09 Sep}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v52/chen16.pdf}, url = {http://proceedings.mlr.press/v52/chen16.html}, abstract = {The space of Bayesian network structures is forbiddingly large and hence numerous techniques have been developed to prune this search space, but without eliminating the optimal structure. Such techniques are critical for structure learning to scale to larger datasets with more variables. Prior works exploited properties of the MDL score to prune away large regions of the search space that can be safely ignored by optimal structure learning algorithms. In this paper, we propose new techniques for pruning regions of the search space that can be safely ignored by algorithms that enumerate the k-best Bayesian network structures. Empirically, we show that these techniques allow a state-of-the-art structure enumeration algorithm to scale to datasets with significantly more variables.} }

%0 Conference Paper %T On Pruning with the MDL Score %A Eunice Yuh-Jie Chen %A Arthur Choi %A Adnan Darwiche %B Proceedings of the Eighth International Conference on Probabilistic Graphical Models %C Proceedings of Machine Learning Research %D 2016 %E Alessandro Antonucci %E Giorgio Corani %E Cassio Polpo Campos} %F pmlr-v52-chen16 %I PMLR %J Proceedings of Machine Learning Research %P 98--109 %U http://proceedings.mlr.press %V 52 %W PMLR %X The space of Bayesian network structures is forbiddingly large and hence numerous techniques have been developed to prune this search space, but without eliminating the optimal structure. Such techniques are critical for structure learning to scale to larger datasets with more variables. Prior works exploited properties of the MDL score to prune away large regions of the search space that can be safely ignored by optimal structure learning algorithms. In this paper, we propose new techniques for pruning regions of the search space that can be safely ignored by algorithms that enumerate the k-best Bayesian network structures. Empirically, we show that these techniques allow a state-of-the-art structure enumeration algorithm to scale to datasets with significantly more variables.

TY - CPAPER TI - On Pruning with the MDL Score AU - Eunice Yuh-Jie Chen AU - Arthur Choi AU - Adnan Darwiche BT - Proceedings of the Eighth International Conference on Probabilistic Graphical Models PY - 2016/08/15 DA - 2016/08/15 ED - Alessandro Antonucci ED - Giorgio Corani ED - Cassio Polpo Campos} ID - pmlr-v52-chen16 PB - PMLR SP - 98 DP - PMLR EP - 109 L1 - http://proceedings.mlr.press/v52/chen16.pdf UR - http://proceedings.mlr.press/v52/chen16.html AB - The space of Bayesian network structures is forbiddingly large and hence numerous techniques have been developed to prune this search space, but without eliminating the optimal structure. Such techniques are critical for structure learning to scale to larger datasets with more variables. Prior works exploited properties of the MDL score to prune away large regions of the search space that can be safely ignored by optimal structure learning algorithms. In this paper, we propose new techniques for pruning regions of the search space that can be safely ignored by algorithms that enumerate the k-best Bayesian network structures. Empirically, we show that these techniques allow a state-of-the-art structure enumeration algorithm to scale to datasets with significantly more variables. ER -