Bayesian Nonlinear Support Vector Machines for Big Data

TL;DR

This paper introduces a scalable Bayesian nonlinear support vector machine method that uses stochastic variational inference and inducing points, offering fast computation, uncertainty quantification, and automatic hyperparameter tuning for large datasets.

Contribution

The paper presents a novel, efficient Bayesian SVM approach that significantly improves scalability and provides uncertainty estimates, surpassing existing Bayesian methods.

Findings

Faster inference compared to competing Bayesian approaches.

Scales easily to millions of data points.

Provides accurate predictive uncertainty estimates.

Abstract

We propose a fast inference method for Bayesian nonlinear support vector machines that leverages stochastic variational inference and inducing points. Our experiments show that the proposed method is faster than competing Bayesian approaches and scales easily to millions of data points. It provides additional features over frequentist competitors such as accurate predictive uncertainty estimates and automatic hyperparameter search.