Temporal Gaussian Process Regression in Logarithmic Time

TL;DR

This paper introduces a parallelization technique for temporal Gaussian process regression that reduces computational complexity from linear to logarithmic time using state-space representation and GPU acceleration.

Contribution

It presents a novel parallelization method that achieves logarithmic time complexity for GP regression, improving efficiency over traditional linear methods.

Findings

Achieves O(log N) time complexity for GP regression

Demonstrates computational benefits on simulated and real datasets

Provides open-source implementation using GPflow

Abstract

The aim of this article is to present a novel parallelization method for temporal Gaussian process (GP) regression problems. The method allows for solving GP regression problems in logarithmic O(log N) time, where N is the number of time steps. Our approach uses the state-space representation of GPs which in its original form allows for linear O(N) time GP regression by leveraging the Kalman filtering and smoothing methods. By using a recently proposed parallelization method for Bayesian filters and smoothers, we are able to reduce the linear computational complexity of the temporal GP regression problems into logarithmic span complexity. This ensures logarithmic time complexity when run on parallel hardware such as a graphics processing unit (GPU). We experimentally demonstrate the computational benefits on simulated and real datasets via our open-source implementation leveraging the GPflow framework.