Mixtures of All Trees
Abstract
Tree-shaped graphical models are widely used for their tractability. However, they unfortunately lack expressive power as they require committing to a particular sparse dependency structure. We propose a novel class of generative models called mixtures of all trees: that is, a mixture over all possible () tree-shaped graphical models over variables. We show that it is possible to parameterize this Mixture of All Trees (MoAT) model compactly (using a polynomial-size representation) in a way that allows for tractable likelihood computation and optimization via stochastic gradient descent. Furthermore, by leveraging the tractability of tree-shaped models, we devise fast-converging conditional sampling algorithms for approximate inference, even though our theoretical analysis suggests that exact computation of marginals in the MoAT model is NP-hard. Empirically, MoAT achieves state-of-the-art performance on density estimation benchmarks when compared against powerful probabilistic models including hidden Chow-Liu Trees.
Keywords
Cite
@article{arxiv.2302.14202,
title = {Mixtures of All Trees},
author = {Nikil Roashan Selvam and Honghua Zhang and Guy Van den Broeck},
journal= {arXiv preprint arXiv:2302.14202},
year = {2023}
}
Comments
Accepted to AISTATS 2023