State Space Gaussian Processes with Non-Gaussian Likelihood

Abstract

We provide a comprehensive overview and tooling for GP modeling with non-Gaussian likelihoods using state space methods. The state space formulation allows for solving one-dimensional GP models in $\mathcal{O}(n)$ time and memory complexity. While existing literature has focused on the connection between GP regression and state space methods, the computational primitives allowing for inference using general likelihoods in combination with the Laplace approximation (LA), variational Bayes (VB), and assumed density filtering (ADF, a.k.a. single-sweep expectation propagation, EP) schemes has been largely overlooked. We present means of combining the efficient $\mathcal{O}(n)$ state space methodology with existing inference methods. We extend existing methods, and provide unifying code implementing all approaches.

References in corpus (4)

• Expectation Propagation for approximate Bayesian inference
• Variational Fourier features for Gaussian processes
• AutoGP: Exploring the Capabilities and Limitations of Gaussian Process Models
• Gaussian Process Kernels for Popular State-Space Time Series Models

Cited by in corpus (7)

• Vecchia-Laplace approximations of generalized Gaussian processes for big non-Gaussian spatial data
• Scalable Exact Inference in Multi-Output Gaussian Processes
• Spatio-Temporal Variational Gaussian Processes
• Bayes-Newton Methods for Approximate Bayesian Inference with PSD Guarantees
• Variational bridge constructs for approximate Gaussian process regression
• Black-Box Inference for Non-Linear Latent Force Models
• Learning Probabilistic Intersection Traffic Models for Trajectory Prediction