Robust Gaussian process regression with G-confluent likelihood
2020 (English)In: IFAC PAPERSONLINE, ELSEVIER , 2020, Vol. 53, no 2, p. 401-406Conference paper, Published paper (Refereed)
Abstract [en]
For robust Gaussian process regression problems where the measurements are contaminated by outliers, a likelihood/measurement noise model with heavy-tailed distributions should be used to improve the prediction performance. In this paper, we propose to use a G-confluent distribution as the measurement noise model and a coordinate ascent variational inference method to infer the overall statistical model. In contrast with the commonly used Students t distribution, the G-confluent distribution can also be written as a Gaussian scale mixture, but its inverse scale follows a Beta distribution rather than a Gamma distribution, and its main advantage is that it is more flexible for modeling outliers while being equally suitable for variational inference. Numerical simulations based on benchmark data show that the G-confluent distribution performs better than or as well as the Students t distribution. Copyright (C) 2020 The Authors.
Place, publisher, year, edition, pages
ELSEVIER , 2020. Vol. 53, no 2, p. 401-406
Keywords [en]
Bayesian methods; Machine learning; Nonparametric methods; Gaussian process regression; Outliers; Variational inference
National Category
Probability Theory and Statistics
Identifiers
URN: urn:nbn:se:liu:diva-176895DOI: 10.1016/j.ifacol.2020.12.197ISI: 000652592500065OAI: oai:DiVA.org:liu-176895DiVA, id: diva2:1572003
Conference
21st IFAC World Congress on Automatic Control - Meeting Societal Challenges, ELECTR NETWORK, jul 11-17, 2020
Note
Funding Agencies|Wallenberg AI, Autonomous Systems and Software Program (wasp) - Knut and Alice Wallenberg Foundation; National Natural Science Foundation of ChinaNational Natural Science Foundation of China (NSFC) [61773329]; Thousand Youth Talents Plan - central government of China; Shenzhen Science and Technology Innovation Council [Ji-20170189 (JCY20170411102101881)]; Chinese University of Hong Kong, Shenzhen [2014.0003.23]; [PF. 01.000249]
2021-06-232021-06-232021-06-23