210

Distributionally Robust Optimization via Ball Oracle Acceleration

Neural Information Processing Systems (NeurIPS), 2022
Abstract

We develop and analyze algorithms for distributionally robust optimization (DRO) of convex losses. In particular, we consider group-structured and bounded ff-divergence uncertainty sets. Our approach relies on an accelerated method that queries a ball optimization oracle, i.e., a subroutine that minimizes the objective within a small ball around the query point. Our main contribution is efficient implementations of this oracle for DRO objectives. For DRO with NN non-smooth loss functions, the resulting algorithms find an ϵ\epsilon-accurate solution with O~(Nϵ2/3+ϵ2)\widetilde{O}\left(N\epsilon^{-2/3} + \epsilon^{-2}\right) first-order oracle queries to individual loss functions. Compared to existing algorithms for this problem, we improve complexity by a factor of up to ϵ4/3\epsilon^{-4/3}.

View on arXiv
Comments on this paper