A Framework for Evaluating Approximation Methods for Gaussian Process Regression
This work addresses the need for better evaluation of approximation methods in Gaussian process regression for machine learning practitioners, but it is incremental as it focuses on assessment rather than introducing new methods.
The paper tackles the problem of evaluating approximation methods for Gaussian process regression, which suffers from high computational costs, by proposing a framework that assesses prediction quality as a function of compute time and comparing it to standard baselines. They empirically test four approximation algorithms on four prediction problems and release code to facilitate future comparisons.
Gaussian process (GP) predictors are an important component of many Bayesian approaches to machine learning. However, even a straightforward implementation of Gaussian process regression (GPR) requires O(n^2) space and O(n^3) time for a dataset of n examples. Several approximation methods have been proposed, but there is a lack of understanding of the relative merits of the different approximations, and in what situations they are most useful. We recommend assessing the quality of the predictions obtained as a function of the compute time taken, and comparing to standard baselines (e.g., Subset of Data and FITC). We empirically investigate four different approximation algorithms on four different prediction problems, and make our code available to encourage future comparisons.