AI RESEARCH

Concave Certificates: Geometric Framework for Distributionally Robust Risk and Complexity Analysis

arXiv CS.LG

ArXi:2601.01311v2 Announce Type: replace-cross Distributionally Robust (DR) optimization aims to certify worst-case risk within a Wasserstein uncertainty set. Current certifications typically rely either on global Lipschitz bounds, which are often conservative, or on local gradient information, which provides only a first-order approximation. This paper