English

Bayesian Robustness: A Nonasymptotic Viewpoint

Machine Learning 2019-07-30 v1 Machine Learning Computation

Abstract

We study the problem of robustly estimating the posterior distribution for the setting where observed data can be contaminated with potentially adversarial outliers. We propose Rob-ULA, a robust variant of the Unadjusted Langevin Algorithm (ULA), and provide a finite-sample analysis of its sampling distribution. In particular, we show that after T=O~(d/εacc)T= \tilde{\mathcal{O}}(d/\varepsilon_{\textsf{acc}}) iterations, we can sample from pTp_T such that dist(pT,p)εacc+O~(ϵ)\text{dist}(p_T, p^*) \leq \varepsilon_{\textsf{acc}} + \tilde{\mathcal{O}}(\epsilon), where ϵ\epsilon is the fraction of corruptions. We corroborate our theoretical analysis with experiments on both synthetic and real-world data sets for mean estimation, regression and binary classification.

Keywords

Cite

@article{arxiv.1907.11826,
  title  = {Bayesian Robustness: A Nonasymptotic Viewpoint},
  author = {Kush Bhatia and Yi-An Ma and Anca D. Dragan and Peter L. Bartlett and Michael I. Jordan},
  journal= {arXiv preprint arXiv:1907.11826},
  year   = {2019}
}

Comments

30 pages, 5 figures

R2 v1 2026-06-23T10:32:30.036Z