Signal Processing and Speech Communication Laboratory
homeresults of the month › Differentiable TAN Structure Learning for Bayesian Network Classifiers

Differentiable TAN Structure Learning for Bayesian Network Classifiers

Published
Fri, Jan 01, 2021
Tags
rotm
Contact
rotm 01 2021

Learning the structure of Bayesian networks is a difficult combinatorial optimization problem. In this paper, we consider learning of tree-augmented naive Bayes (TAN) structures for Bayesian network classifiers with discrete input features. Instead of performing a combinatorial optimization over the space of possible graph structures, the proposed method learns a distribution over graph structures. After training, we select the most probable structure of this distribution. This allows for a joint training of the Bayesian network parameters along with its TAN structure using gradient-based optimization. The proposed method is agnostic to the specific loss and only requires that it is differentiable. We perform extensive experiments using a hybrid generative-discriminative loss based on the discriminative probabilistic margin. Our method consistently outperforms random TAN structures and Chow-Liu TAN structures.

The paper was presented at the International Conference on Probabilistic Graphical Models (PGM 2020) and can be found at https://arxiv.org/abs/2008.09566. Code for the experiments is available at https://github.com/wroth8/bnc.

Figure: Left: Our method learns a distribution over the graph structure. Connections that are more probable are drawn bold. Right: After training, we select the most probable structure, i.e., we select the most probable parent of each node.

Browse the Results of the Month archive.