Adversarial Risk via Optimal Transport and Optimal Couplings

DOI10.1109/TIT.2021.3100107MaRDI QIDQ4958283zbMATH OpenOpenAlexFDO

Authors Muni Sreenivas Pydi, Varun Jog

Publication date 7 September 2021

Published in IEEE Transactions on Information Theory (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1912.02794

zbMATH Keywords

adversarial classifiersadversarial riskoptimal transport plans for univariate distributions

Mathematics Subject Classification ID

Classification and discrimination; cluster analysis (statistical aspects) (62H30) Information theory (general) (94A15)

Abstract: Modern machine learning algorithms perform poorly on adversarially manipulated data. Adversarial risk quantifies the error of classifiers in adversarial settings; adversarial classifiers minimize adversarial risk. In this paper, we analyze adversarial risk and adversarial classifiers from an optimal transport perspective. We show that the optimal adversarial risk for binary classification with 0-1 loss is determined by an optimal transport cost between the probability distributions of the two classes. We develop optimal transport plans (probabilistic couplings) for univariate distributions such as the normal, the uniform, and the triangular distribution. We also derive optimal adversarial classifiers in these settings. Our analysis leads to algorithm-independent fundamental limits on adversarial risk, which we calculate for several real-world datasets. We extend our results to general loss functions under convexity and smoothness assumptions.

Cited in

(9)

• Achieving optimal adversarial accuracy for adversarial deep learning using Stackelberg games
• Precise statistical analysis of classification accuracies for adversarial training
• Adversarial classification via distributional robustness with Wasserstein ambiguity
• Robust Classification Under \ell₀ Attack for the Gaussian Mixture Model
• Nonasymptotic bounds for adversarial excess risk under misspecified models
• Uniform convergence of adversarially robust classifiers
• On the existence of solutions to adversarial training in multiclass classification
• Distributionally robust optimization
• An optimal transport approach for computing adversarial training lower bounds in multiclass classification

This page was built for publication: Adversarial Risk via Optimal Transport and Optimal Couplings

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4958283)