Non-Gaussian Discriminative Factor Models via the Max-Margin Rank-Likelihood

TL;DR

This paper introduces a novel Bayesian discriminative factor model that leverages a max-margin rank-likelihood for non-Gaussian data, combining linear and nonlinear classifiers with efficient inference methods, showing superior results in biological data analysis.

Contribution

It proposes a new max-margin rank-likelihood and integrates it with Bayesian support vector machines for discriminative factor analysis, extending to nonlinear models with Dirichlet process mixtures.

Findings

Superior performance on benchmark datasets

Effective in computational biology applications

Efficient inference via MCMC and variational Bayes

Abstract

We consider the problem of discriminative factor analysis for data that are in general non-Gaussian. A Bayesian model based on the ranks of the data is proposed. We first introduce a new {\em max-margin} version of the rank-likelihood. A discriminative factor model is then developed, integrating the max-margin rank-likelihood and (linear) Bayesian support vector machines, which are also built on the max-margin principle. The discriminative factor model is further extended to the {\em nonlinear} case through mixtures of local linear classifiers, via Dirichlet processes. Fully local conjugacy of the model yields efficient inference with both Markov Chain Monte Carlo and variational Bayes approaches. Extensive experiments on benchmark and real data demonstrate superior performance of the proposed model and its potential for applications in computational biology.