Deep Structured Mixtures of Gaussian Processes
This work addresses the problem of scaling GPs for practitioners in machine learning, offering a principled and efficient alternative to existing approximations.
The paper tackles the scalability and approximation limitations of Gaussian Processes (GPs) by introducing deep structured mixtures of GP experts, which enable exact posterior inference with lower computational costs and better capture predictive uncertainties than previous methods, achieving competitive or superior performance in experiments.
Gaussian Processes (GPs) are powerful non-parametric Bayesian regression models that allow exact posterior inference, but exhibit high computational and memory costs. In order to improve scalability of GPs, approximate posterior inference is frequently employed, where a prominent class of approximation techniques is based on local GP experts. However, local-expert techniques proposed so far are either not well-principled, come with limited approximation guarantees, or lead to intractable models. In this paper, we introduce deep structured mixtures of GP experts, a stochastic process model which i) allows exact posterior inference, ii) has attractive computational and memory costs, and iii) when used as GP approximation, captures predictive uncertainties consistently better than previous expert-based approximations. In a variety of experiments, we show that deep structured mixtures have a low approximation error and often perform competitive or outperform prior work.