Current browse context:
math.OC
Change to browse by:
References & Citations
Mathematics > Optimization and Control
Title: Distributionally Robust Gaussian Process Regression and Bayesian Inverse Problems
(Submitted on 26 May 2022 (v1), last revised 20 Oct 2022 (this version, v4))
Abstract: We study a distributionally robust optimization formulation (i.e., a min-max game) for two representative problems in Bayesian nonparametric estimation: Gaussian process regression and, more generally, linear inverse problems. Our formulation seeks the best mean-squared error predictor, in an infinite-dimensional space, against an adversary who chooses the worst-case model in a Wasserstein ball around a nominal infinite-dimensional Bayesian model. The transport cost is chosen to control features such as the degree of roughness of the sample paths that the adversary is allowed to inject. We show that the game has a well-defined value (i.e., strong duality holds in the sense that max-min equals min-max) and that there exists a unique Nash equilibrium which can be computed by a sequence of finite-dimensional approximations. Crucially, the worst-case distribution is itself Gaussian. We explore properties of the Nash equilibrium and the effects of hyperparameters through a set of numerical experiments, demonstrating the versatility of our modeling framework.
Submission history
From: Xuhui Zhang [view email][v1] Thu, 26 May 2022 02:34:08 GMT (12466kb,D)
[v2] Thu, 9 Jun 2022 17:09:31 GMT (12466kb,D)
[v3] Tue, 21 Jun 2022 23:18:12 GMT (12466kb,D)
[v4] Thu, 20 Oct 2022 05:52:18 GMT (9598kb,D)
Link back to: arXiv, form interface, contact.