On adversarial robustness and the use of Wasserstein ascent-descent dynamics to enforce it

DOI10.1093/IMAIAI/IAAE018MaRDI QIDQ6638523zbMATH OpenFDO

Authors Camilo Andrés García Trillos, Nicolás García Trillos

Publication date 14 November 2024

Published in Information and Inference: A Journal of the IMA (Search for Journal in Brave)

Full work available at URL https://doi.org/10.1093/IMAIAI/IAAE018

zbMATH Keywords

adversarial robustnessadversarial traininginteracting particle systemmean-field limitminmax gamesNash equilibriumWasserstein gradient flowWasserstein-Fisher-Rao metric

Mathematics Subject Classification ID

PDEs in connection with mean field game theory (35Q89) Learning and adaptive systems in artificial intelligence (68T05) Minimax problems in mathematical programming (90C47) Applications of mathematical programming (90C90) Mean field games (aspects of game theory) (91A16)

Recommendations

• A fast saddle-point dynamical system approach to robust deep learning
• Adversarial classification via distributional robustness with Wasserstein ambiguity
• Optimal training for adversarial games
• The geometry of adversarial training in binary classification
• Two steps at a time-taking GAN training in stride with Tseng's method

Cited in

(1)

• The geometry of adversarial training in binary classification

This page was built for publication: On adversarial robustness and the use of Wasserstein ascent-descent dynamics to enforce it

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6638523)