Neural Graph Revealers

Sparse graph recovery methods work well where the data follows their assumptions, however, they are not always designed for doing downstream probabilistic queries. This limits their adoption to only identifying connections among domain variables. On the other hand, Probabilistic Graphical Models (PGMs) learn an underlying base graph together with a distribution over the variables (nodes). PGM design choices are carefully made such that the inference and sampling algorithms are efficient. This results in certain restrictions and simplifying assumptions. In this work, we propose Neural Graph Revealers (NGRs) which attempt to efficiently merge the sparse graph recovery methods with PGMs into a single flow. The task is to recover a sparse graph showing connections between the features and learn a probability distribution over them at the same time. NGRs use a neural network as a multi-task learning framework. We introduce graph-constrained path norm that NGRs leverage to learn a graphical model that captures complex non-linear functional dependencies between features in the form of an undirected sparse graph. NGRs can handle multimodal inputs like images, text, categorical data, embeddings etc. which are not straightforward to incorporate in the existing methods. We show experimental results on data from Gaussian graphical models and a multimodal infant mortality dataset by CDC (Software: https://github.com/harshs27/neural-graph-revealers).