GPy-ABCD: A Configurable Automatic Bayesian Covariance Discovery Implementation
Thomas Fletcher, Alan Bundy, Kwabena Nuamah · Edinburgh Research Explorer (University of Edinburgh) · 2021
Gaussian Processes (GPs) are a very flexible class of nonparametric models which are able to fit data with very few assumptions, namely just the type of correlation (kernel) the data is expected to display. Automatic Bayesian Covariance Discovery (ABCD)1 is an iterative modular Gaussian Process regression framework aimed at removing the requirement for even this initial correlation form assumption. GPy-ABCD2 is a new implementation of an ABCD system built for ease of use and configurability; it can produce short text descriptions of fit models, it uses a revised model-space search algorithm and it removes a search bias which was required in order to retain model explainability in the original system.