Sparse Additive Gaussian Process Regression

• Activity Number: 567 - New Approaches for Sparse Gaussian Processes
• Type: Invited
• Date/Time: Thursday, August 6, 2020 : 3:00 PM to 4:50 PM
• Sponsor: Uncertainty Quantification in Complex Systems Interest Group
• Author(s): Hengrui Luo and Giovanni Nattino and Matthew Pratola*
• Companies: The Ohio State University and L'Istituto di Ricerche Farmacologiche Mario Negri IRCCS desidera and The Ohio State University
• Keywords: Sparse Gaussian Process; Recursive Partition Scheme; Bayesian Additive Model; Nonparametric Regression

Abstract

We introduce a novel model for Gaussian process (GP) regression in the fully Bayesian setting. Motivated by the idea of sparsification, localization and Bayesian additive modeling, our model is built around a recursive partitioning (RP) scheme. Within each RP partition, a sparse GP regression model is fitted. A Bayesian additive framework combines the partitions, allowing the model to admit both global trends and local refinements on which a sparse GP construction enables efficient computation. The model addresses both the problem of efficiency in fitting a full Gaussian process regression model and the problem of prediction performance associated with a single sparse Gaussian process.