Macau: Scalable Bayesian Multi-relational Factorization with Side Information using MCMC

TL;DR

Macau is a scalable Bayesian factorization method that effectively models heterogeneous data, incorporates side information, and handles large-scale, sparse relational datasets with high efficiency.

Contribution

The paper introduces Macau, a novel Bayesian factorization approach that scales to large, sparse multi-relational data and integrates side information using a specialized MCMC sampling technique.

Findings

Successfully applied to drug-protein activity prediction

Handles millions of entities and relations efficiently

Incorporates side information for improved predictions

Abstract

We propose Macau, a powerful and flexible Bayesian factorization method for heterogeneous data. Our model can factorize any set of entities and relations that can be represented by a relational model, including tensors and also multiple relations for each entity. Macau can also incorporate side information, specifically entity and relation features, which are crucial for predicting sparsely observed relations. Macau scales to millions of entity instances, hundred millions of observations, and sparse entity features with millions of dimensions. To achieve the scale up, we specially designed sampling procedure for entity and relation features that relies primarily on noise injection in linear regressions. We show performance and advanced features of Macau in a set of experiments, including challenging drug-protein activity prediction task.