Scaling Up Bayesian DAG Sampling

Daniele Nikzad, Alexander Zhilkin, Juha Harviainen, Jack Kuipers, Giusi Moffa, Mikko Koivisto
Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence, PMLR 337:4949-4971, 2026.

Abstract

{Bayesian} inference of {Bayesian} network structures is often performed by sampling directed acyclic graphs along an appropriately constructed {Markov} chain. We present two techniques to improve sampling. First, we give an efficient implementation of basic moves, which add, delete, or reverse a single arc. Second, we expedite summing over parent sets, an expensive task required for more sophisticated moves: we devise a preprocessing method to prune possible parent sets so as to approximately preserve the sums. Our empirical study shows that our techniques can yield substantial efficiency gains compared to previous methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v337-nikzad26a, title = {Scaling Up {Bayesian} {DAG} Sampling}, author = {Nikzad, Daniele and Zhilkin, Alexander and Harviainen, Juha and Kuipers, Jack and Moffa, Giusi and Koivisto, Mikko}, booktitle = {Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence}, pages = {4949--4971}, year = {2026}, editor = {Perković, Emilija and Malinsky, Daniel}, volume = {337}, series = {Proceedings of Machine Learning Research}, month = {17--21 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v337/main/assets/nikzad26a/nikzad26a.pdf}, url = {https://proceedings.mlr.press/v337/nikzad26a.html}, abstract = {{Bayesian} inference of {Bayesian} network structures is often performed by sampling directed acyclic graphs along an appropriately constructed {Markov} chain. We present two techniques to improve sampling. First, we give an efficient implementation of basic moves, which add, delete, or reverse a single arc. Second, we expedite summing over parent sets, an expensive task required for more sophisticated moves: we devise a preprocessing method to prune possible parent sets so as to approximately preserve the sums. Our empirical study shows that our techniques can yield substantial efficiency gains compared to previous methods.} }
Endnote
%0 Conference Paper %T Scaling Up Bayesian DAG Sampling %A Daniele Nikzad %A Alexander Zhilkin %A Juha Harviainen %A Jack Kuipers %A Giusi Moffa %A Mikko Koivisto %B Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2026 %E Emilija Perković %E Daniel Malinsky %F pmlr-v337-nikzad26a %I PMLR %P 4949--4971 %U https://proceedings.mlr.press/v337/nikzad26a.html %V 337 %X {Bayesian} inference of {Bayesian} network structures is often performed by sampling directed acyclic graphs along an appropriately constructed {Markov} chain. We present two techniques to improve sampling. First, we give an efficient implementation of basic moves, which add, delete, or reverse a single arc. Second, we expedite summing over parent sets, an expensive task required for more sophisticated moves: we devise a preprocessing method to prune possible parent sets so as to approximately preserve the sums. Our empirical study shows that our techniques can yield substantial efficiency gains compared to previous methods.
APA
Nikzad, D., Zhilkin, A., Harviainen, J., Kuipers, J., Moffa, G. & Koivisto, M.. (2026). Scaling Up Bayesian DAG Sampling. Proceedings of the 42nd Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research 337:4949-4971 Available from https://proceedings.mlr.press/v337/nikzad26a.html.

Related Material