Rates of Convergence for Sparse Variational Gaussian Process Regression

Abstract

Excellent variational approximations to Gaussian process posteriors have been developed which avoid the $\mathcal{O}\left(N^3\right)$ scaling with dataset size $N$. They reduce the computational cost to $\mathcal{O}\left(N{M}^2\right)$, with $M\ll N$ being the number of inducing variables, which summarise the process. While the computational cost seems to be linear in $N$, the true complexity of the algorithm depends on how $M$ must increase to ensure a certain quality of approximation. We address this by characterising the behavior of an upper bound on the KL divergence to the posterior. We show that with high probability the KL divergence can be made arbitrarily small by growing $M$ more slowly than $N$. A particular case of interest is that for regression with normally distributed inputs in D-dimensions with the popular Squared Exponential kernel, $M=\mathcal{O}({\log }^{D}N)$ is sufficient. Our results show that as datasets grow, Gaussian process posteriors can truly be approximated cheaply, and provide a concrete rule for how to increase $M$ in continual learning scenarios.

International Conference on Machine Learning (ICML 2019)

References in corpus (6)

• Gaussian Processes for Big Data
• Scalable Variational Gaussian Process Classification
• Variational Fourier features for Gaussian processes
• Fast DPP Sampling for Nyström with Application to Kernel Methods
• Exact sampling of determinantal point processes with sublinear time preprocessing
• Exact Sampling from Determinantal Point Processes

Cited by in corpus (12)

• On Feature Collapse and Deep Kernel Learning for Single Forward Pass Uncertainty
• Variational Inference of Joint Models using Multivariate Gaussian Convolution Processes
• Machine Learning based optimization for interval uncertainty propagation
• Scalable Gaussian Process Variational Autoencoders
• Convergence Guarantees for Gaussian Process Means With Misspecified Likelihoods and Smoothness
• Scalable Gaussian Processes on Discrete Domains
• Deep Bayesian Gaussian Processes for Uncertainty Estimation in Electronic Health Records
• Iterative Correction of Sensor Degradation and a Bayesian Multi-Sensor Data Fusion Method
• Longitudinal Deep Kernel Gaussian Process Regression
• The Renyi Gaussian Process: Towards Improved Generalization
• Gaussian Experts Selection using Graphical Models
• Scalable Bayesian Optimization with Sparse Gaussian Process Models