Scaling Multidimensional Inference for Structured Gaussian Processes

Structured Variational Inference for Coupled Gaussian Processes

Sparse variational approximations allow for principled and scalable inference in Gaussian Process (GP) models. In settings where several GPs are part of the generative model, theses GPs are a posteriori coupled. For many applications such as regression where predictive accuracy is the quantity of interest, this coupling is not crucial. Howewer if one is interested in posterior uncertainty, it c...

Structured Variational Inference for Coupled Gaussian Processes

Sparse variational approximations allow for principled and scalable inference in Gaussian Process (GP) models. In settings where several GPs are part of the generative model, these GPs are a posteriori coupled. For many applications such as regression where predictive accuracy is the quantity of interest, this coupling is not crucial. Howewer if one is interested in posterior uncertainty, it ca...

Scaling Multidimensional Gaussian Processes using Projected Additive Approximations

Exact Gaussian Process (GP) regression has O(N) runtime for data size N , making it intractable for large N . Advances in GP scaling have not been extended to the multidimensional input setting, despite the preponderance of multidimensional applications. This paper introduces and tests a novel method of projected additive approximation to multidimensional GPs. We illustrate the power of this me...

Scalable Inference for Structured Gaussian Process Models

The generic inference and learning algorithm for Gaussian Process (GP) regression has O(N3) runtime and O(N2) memory complexity, where N is the number of observations in the dataset. Given the computational resources available to a present-day workstation, this implies that GP regression simply cannot be run on large datasets. The need to use nonGaussian likelihood functions for tasks such as c...

Multidimensional Wick-itô Formula for Gaussian Processes