English

Safe Convex Learning under Uncertain Constraints

Optimization and Control 2019-12-10 v2

Abstract

We address the problem of minimizing a convex smooth function f(x)f(x) over a compact polyhedral set DD given a stochastic zeroth-order constraint feedback model. This problem arises in safety-critical machine learning applications, such as personalized medicine and robotics. In such cases, one needs to ensure constraints are satisfied while exploring the decision space to find optimum of the loss function. We propose a new variant of the Frank-Wolfe algorithm, which applies to the case of uncertain linear constraints. Using robust optimization, we provide the convergence rate of the algorithm while guaranteeing feasibility of all iterates, with high probability.

Keywords

Cite

@article{arxiv.1903.04626,
  title  = {Safe Convex Learning under Uncertain Constraints},
  author = {Ilnura Usmanova and Andreas Krause and Maryam Kamgarpour},
  journal= {arXiv preprint arXiv:1903.04626},
  year   = {2019}
}

Comments

15 pages, 7 figures, AISTATS 2019

R2 v1 2026-06-23T08:04:57.554Z