Edinburgh Research Explorer

Bayesian Multitask Classification With Gaussian Process Priors

Research output: Contribution to journalArticle

Original languageEnglish
Pages (from-to)2011-2021
Number of pages11
JournalIEEE Transactions on Neural Networks
Volume22
Issue number12
DOIs
Publication statusPublished - 1 Dec 2011

Abstract

We present a novel approach to multitask learning in classification problems based on Gaussian process (GP) classification. The method extends previous work on multitask GP regression, constraining the overall covariance (across tasks and data points) to factorize as a Kronecker product. Fully Bayesian inference is possible but time consuming using sampling techniques. We propose approximations based on the popular variational Bayes and expectation propagation frameworks, showing that they both achieve excellent accuracy when compared to Gibbs sampling, in a fraction of time. We present results on a toy dataset and two real datasets, showing improved performance against the baseline results obtained by learning each task independently. We also compare with a recently proposed state-of-the-art approach based on support vector machines, obtaining comparable or better results.

    Research areas

  • Bayes methods, Gaussian processes, pattern classification, regression analysis, variational techniques, Bayesian inference, Bayesian multitask classification, Gaussian process classification, Gaussian process priors, Gibbs sampling, Kronecker product, classification problem, expectation propagation frameworks, multitask GP regression, multitask learning, sampling technique, support vector machines, variational Bayes, Approximation methods, Bayesian methods, Correlation, Covariance matrix, Optimization, Training, classification, Algorithms, Artificial Intelligence, Bayes Theorem, Computer Simulation, Models, Statistical, Normal Distribution, Pattern Recognition, Automated

ID: 21808876