Max-Margin Nonparametric Latent Feature Models for Link Prediction

TL;DR

This paper introduces a max-margin nonparametric latent feature model that combines max-margin learning with Bayesian nonparametrics for improved link prediction and social dimension inference, demonstrating efficiency and effectiveness.

Contribution

It proposes a novel model uniting max-margin learning with Bayesian nonparametrics, enabling automatic discovery of discriminative latent features for link prediction.

Findings

Efficient posterior inference without nonlinear link likelihood.

Automatic inference of latent social dimensions.

Improved link prediction performance on real datasets.

Abstract

We present a max-margin nonparametric latent feature model, which unites the ideas of max-margin learning and Bayesian nonparametrics to discover discriminative latent features for link prediction and automatically infer the unknown latent social dimension. By minimizing a hinge-loss using the linear expectation operator, we can perform posterior inference efficiently without dealing with a highly nonlinear link likelihood function; by using a fully-Bayesian formulation, we can avoid tuning regularization constants. Experimental results on real datasets appear to demonstrate the benefits inherited from max-margin learning and fully-Bayesian nonparametric inference.