Current browse context:
cs.LG
Change to browse by:
References & Citations
Computer Science > Machine Learning
Title: The Geometry of Adversarial Training in Binary Classification
(Submitted on 26 Nov 2021 (v1), last revised 1 Aug 2022 (this version, v2))
Abstract: We establish an equivalence between a family of adversarial training problems for non-parametric binary classification and a family of regularized risk minimization problems where the regularizer is a nonlocal perimeter functional. The resulting regularized risk minimization problems admit exact convex relaxations of the type $L^1+$ (nonlocal) $\operatorname{TV}$, a form frequently studied in image analysis and graph-based learning. A rich geometric structure is revealed by this reformulation which in turn allows us to establish a series of properties of optimal solutions of the original problem, including the existence of minimal and maximal solutions (interpreted in a suitable sense), and the existence of regular solutions (also interpreted in a suitable sense). In addition, we highlight how the connection between adversarial training and perimeter minimization problems provides a novel, directly interpretable, statistical motivation for a family of regularized risk minimization problems involving perimeter/total variation. The majority of our theoretical results are independent of the distance used to define adversarial attacks.
Submission history
From: Leon Bungert [view email][v1] Fri, 26 Nov 2021 17:19:50 GMT (492kb,D)
[v2] Mon, 1 Aug 2022 08:16:49 GMT (2074kb,D)
Link back to: arXiv, form interface, contact.