Probabilistic DAG search
Julia Grosse, Cheng Zhang, Philipp Hennig
Abstract
Exciting contemporary machine learning problems have recently been phrased in the classic formalism of tree search — most famously, the game of Go. Interestingly, the state-space underlying these sequential decision-making problems often posses a more general latent structure than can be captured by a tree. In this work, we develop a probabilistic framework to exploit a search space’s latent structure and thereby share information across the search tree. The method is based on a combination of approximate inference in jointly Gaussian models for the explored part of the problem, and an abstraction for the unexplored part that imposes a reduction of complexity ad hoc. We empirically find our algorithm to compare favorably to existing non-probabilistic alternatives in Tic-Tac-Toe and a feature selection application.
BibTeX
@InProceedings{pmlr-v161-grosse21a,
title = {Probabilistic DAG search},
author = {Grosse, Julia and Zhang, Cheng and Hennig, Philipp},
booktitle = {Proceedings of the Thirty-Seventh Conference on Uncertainty in Artificial Intelligence},
pages = {1424--1433},
year = {2021},
editor = {de Campos, Cassio and Maathuis, Marloes H.},
volume = {161},
series = {Proceedings of Machine Learning Research},
month = {27--30 Jul},
publisher = {PMLR},
pdf = {https://proceedings.mlr.press/v161/grosse21a/grosse21a.pdf},
url = {https://proceedings.mlr.press/v161/grosse21a.html},
abstract = {Exciting contemporary machine learning problems have recently been phrased in the classic formalism of tree search — most famously, the game of Go. Interestingly, the state-space underlying these sequential decision-making problems often posses a more general latent structure than can be captured by a tree. In this work, we develop a probabilistic framework to exploit a search space’s latent structure and thereby share information across the search tree. The method is based on a combination of approximate inference in jointly Gaussian models for the explored part of the problem, and an abstraction for the unexplored part that imposes a reduction of complexity ad hoc. We empirically find our algorithm to compare favorably to existing non-probabilistic alternatives in Tic-Tac-Toe and a feature selection application.}
}