Adversarial classification via distributional robustness with Wasserstein ambiguity

DOI10.1007/s10107-022-01796-6OpenAlexW3031808047MaRDI QIDQ2693647

Stephen J. Wright, Nam Ho-Nguyen

Publication date: 24 March 2023

Published in: Mathematical Programming. Series A. Series B (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/2005.13815

zbMATH Keywords

nonconvexdistributional robustnessadversarial classificationmarginWasserstein ambiguityramp loss

Mathematics Subject Classification ID

Nonconvex programming, global optimization (90C26) Nonlinear programming (90C30) Computational aspects of data analysis and big data (68T09) Robustness in mathematical programming (90C17)

Related Items

On approximations of data-driven chance constrained programs over Wasserstein balls

Uses Software

• L-BFGS
• DeepFool

Cites Work

• Unnamed Item
• Unnamed Item
• Unnamed Item
• Unnamed Item
• Characterization of the equivalence of robustification and regularization in linear and matrix regression
• On the limited memory BFGS method for large scale optimization
• Analysis of classifiers' robustness to adversarial perturbations
• Data-driven distributionally robust optimization using the Wasserstein metric: performance guarantees and tractable reformulations
• Shrinkage estimation
• On distributionally robust chance constrained programs with Wasserstein distance
• On handling indicator constraints in mixed integer programming
• Smoothing and First Order Methods: A Unified Framework
• Support Vector Machines with the Ramp Loss and the Hard Margin Loss
• Oracle-Based Robust Optimization via Online Learning
• Robust Truncated Hinge Loss Support Vector Machines
• Cutting-set methods for robust convex optimization with pessimizing oracles
• On ψ-Learning
• Adversarial Risk via Optimal Transport and Optimal Couplings
• Online First-Order Framework for Robust Convex Optimization
• Regularization via Mass Transportation
• Quantifying Distributional Model Risk via Optimal Transport
• Nonconvex Optimization Meets Low-Rank Matrix Factorization: An Overview
• Linear and Nonlinear Separation of Patterns by Linear Programming
• AN OLD‐NEW CONCEPT OF CONVEX RISK MEASURES: THE OPTIMIZED CERTAINTY EQUIVALENT
• Exact and inexact subsampled Newton methods for optimization