Curriculum Learning of Bayesian Network Structures

Yanpeng Zhao, Yetian Chen, Kewei Tu, Jin Tian
Asian Conference on Machine Learning, PMLR 45:269-284, 2016.

Abstract

Bayesian networks (BNs) are directed graphical models that have been widely used in various tasks for probabilistic reasoning and causal modeling. One major challenge in these tasks is to learn the BN structures from data. In this paper, we propose a novel heuristic algorithm for BN structure learning that takes advantage of the idea of \emphcurriculum learning. Our algorithm learns the BN structure by stages. At each stage a subnet is learned over a selected subset of the random variables conditioned on fixed values of the rest of the variables. The selected subset grows with stages and eventually includes all the variables. We prove theoretical advantages of our algorithm and also empirically show that it outperformed the state-of-the-art heuristic approach in learning BN structures.

Cite this Paper

BibTeX
@InProceedings{pmlr-v45-Zhao15a, title = {Curriculum Learning of Bayesian Network Structures}, author = {Zhao, Yanpeng and Chen, Yetian and Tu, Kewei and Tian, Jin}, booktitle = {Asian Conference on Machine Learning}, pages = {269--284}, year = {2016}, editor = {Holmes, Geoffrey and Liu, Tie-Yan}, volume = {45}, series = {Proceedings of Machine Learning Research}, address = {Hong Kong}, month = {20--22 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v45/Zhao15a.pdf}, url = {https://proceedings.mlr.press/v45/Zhao15a.html}, abstract = {Bayesian networks (BNs) are directed graphical models that have been widely used in various tasks for probabilistic reasoning and causal modeling. One major challenge in these tasks is to learn the BN structures from data. In this paper, we propose a novel heuristic algorithm for BN structure learning that takes advantage of the idea of \emphcurriculum learning. Our algorithm learns the BN structure by stages. At each stage a subnet is learned over a selected subset of the random variables conditioned on fixed values of the rest of the variables. The selected subset grows with stages and eventually includes all the variables. We prove theoretical advantages of our algorithm and also empirically show that it outperformed the state-of-the-art heuristic approach in learning BN structures. } }
Endnote
%0 Conference Paper %T Curriculum Learning of Bayesian Network Structures %A Yanpeng Zhao %A Yetian Chen %A Kewei Tu %A Jin Tian %B Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2016 %E Geoffrey Holmes %E Tie-Yan Liu %F pmlr-v45-Zhao15a %I PMLR %P 269--284 %U https://proceedings.mlr.press/v45/Zhao15a.html %V 45 %X Bayesian networks (BNs) are directed graphical models that have been widely used in various tasks for probabilistic reasoning and causal modeling. One major challenge in these tasks is to learn the BN structures from data. In this paper, we propose a novel heuristic algorithm for BN structure learning that takes advantage of the idea of \emphcurriculum learning. Our algorithm learns the BN structure by stages. At each stage a subnet is learned over a selected subset of the random variables conditioned on fixed values of the rest of the variables. The selected subset grows with stages and eventually includes all the variables. We prove theoretical advantages of our algorithm and also empirically show that it outperformed the state-of-the-art heuristic approach in learning BN structures.
RIS
TY - CPAPER TI - Curriculum Learning of Bayesian Network Structures AU - Yanpeng Zhao AU - Yetian Chen AU - Kewei Tu AU - Jin Tian BT - Asian Conference on Machine Learning DA - 2016/02/25 ED - Geoffrey Holmes ED - Tie-Yan Liu ID - pmlr-v45-Zhao15a PB - PMLR DP - Proceedings of Machine Learning Research VL - 45 SP - 269 EP - 284 L1 - http://proceedings.mlr.press/v45/Zhao15a.pdf UR - https://proceedings.mlr.press/v45/Zhao15a.html AB - Bayesian networks (BNs) are directed graphical models that have been widely used in various tasks for probabilistic reasoning and causal modeling. One major challenge in these tasks is to learn the BN structures from data. In this paper, we propose a novel heuristic algorithm for BN structure learning that takes advantage of the idea of \emphcurriculum learning. Our algorithm learns the BN structure by stages. At each stage a subnet is learned over a selected subset of the random variables conditioned on fixed values of the rest of the variables. The selected subset grows with stages and eventually includes all the variables. We prove theoretical advantages of our algorithm and also empirically show that it outperformed the state-of-the-art heuristic approach in learning BN structures. ER -
APA
Zhao, Y., Chen, Y., Tu, K. & Tian, J.. (2016). Curriculum Learning of Bayesian Network Structures. Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 45:269-284 Available from https://proceedings.mlr.press/v45/Zhao15a.html.

Related Material