Hinge-loss Markov Random Fields: Convex Inference for Structured Prediction

TL;DR

This paper introduces hinge-loss Markov random fields (HL-MRFs), a class of continuous graphical models that enable fast, scalable, and accurate structured prediction, outperforming some discrete models in various applications.

Contribution

The paper presents the first scalable inference algorithm for HL-MRFs and demonstrates their effectiveness across multiple structured prediction tasks.

Findings

HL-MRFs match or surpass state-of-the-art methods in predictive performance.

The proposed inference algorithm is scalable to the full class of HL-MRFs.

HL-MRFs effectively model confidences in discrete predictions using continuous variables.

Abstract

Graphical models for structured domains are powerful tools, but the computational complexities of combinatorial prediction spaces can force restrictions on models, or require approximate inference in order to be tractable. Instead of working in a combinatorial space, we use hinge-loss Markov random fields (HL-MRFs), an expressive class of graphical models with log-concave density functions over continuous variables, which can represent confidences in discrete predictions. This paper demonstrates that HL-MRFs are general tools for fast and accurate structured prediction. We introduce the first inference algorithm that is both scalable and applicable to the full class of HL-MRFs, and show how to train HL-MRFs with several learning algorithms. Our experiments show that HL-MRFs match or surpass the predictive performance of state-of-the-art methods, including discrete models, in four application domains.