Skip to content
conftrace_
2005 JMLR JMLR 2005

Assessing Approximate Inference for Binary Gaussian Process Classification

Abstract

Gaussian process priors can be used to define flexible, probabilistic classification models. Unfortunately exact Bayesian inference is analytically intractable and various approximation techniques have been proposed. In this work we review and compare Laplace's method and Expectation Propagation for approximate Bayesian inference in the binary Gaussian process classification model. We present a comprehensive comparison of the approximations, their predictive performance and marginal likelihood estimates to results obtained by MCMC sampling. We explain theoretically and corroborate empirically the advantages of Expectation Propagation compared to Laplace's method. [abs] [ pdf ][ bib ] © JMLR 2005. (edit, beta)

📈 Trend Setter - Variational Inference
🧭 Keyword Pioneer - predictive performance
🐣 Hot Topic Early Bird - gaussian process

Authors

Malte Kuss, Carl Edward Rasmussen

Topics

Machine Learning > Core Methods > Classification Machine Learning > Optimization & Theory > Bayesian Inference Machine Learning > Bayesian & Probabilistic > Variational Inference Machine Learning > Bayesian & Probabilistic > Gaussian Processes

Keywords

approximate inference bayesian inference markov chain monte carlo laplace approximation gaussian process expectation propagation gaussian process classification predictive performance
Download PDF

More from JMLR 2005

All JMLR 2005 papers →
Fast Kernel Classifiers with Online and Active Learning
Managing Diversity in Regression Ensembles
Active Coevolutionary Learning of Deterministic Finite Automata
Clustering with Bregman Divergences
Combining Information Extraction Systems Using Voting and Stacked Generalization
Probabilistic Non-linear Principal Component Analysis with Gaussian Process Latent Variable Models