TY - GEN
T1 - Deep Structured Mixtures of Gaussian Processes
AU - Trapp, Martin
AU - Peharz, Robert
AU - Pernkopf, Franz
AU - Rasmussen, Carl E.
PY - 2020/4/26
Y1 - 2020/4/26
N2 - 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.
AB - 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.
KW - cs.LG
KW - stat.ML
UR - http://www.scopus.com/inward/record.url?scp=85161860408&partnerID=8YFLogxK
M3 - Conference contribution
T3 - Proceedings of Machine Learning Research (PMLR)
SP - 2251
EP - 2261
BT - Proceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics (PMLR)
A2 - Chiappa, Silvia
A2 - Calandra, Roberta
T2 - 23rd International Conference on Artificial Intelligence and Statistics, ONLINE
Y2 - 26 August 2020 through 28 August 2020
ER -