Achieving optimal adversarial accuracy for adversarial deep learning using Stackelberg games

DOI10.1007/S10473-022-0613-YMaRDI QIDQ2080981zbMATH OpenOpenAlexFDO

Authors Shuang Liu, Lijia Yu, Xiao-Shan Gao

Publication date 12 October 2022

Published in Acta Mathematica Scientia. Series B. (English Edition) (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/2207.08137

zbMATH Keywords

adversarial accuracyadversarial deep learningoptimal robust DNNStackelberg gametrade-off resultuniversal adversarial attack

Mathematics Subject Classification ID

Robustness and adaptive procedures (parametric inference) (62F35) Learning and adaptive systems in artificial intelligence (68T05) Applications of game theory (91A80)

Recommendations

• Gradient methods for solving Stackelberg games
• Scalable Optimal Classifiers for Adversarial Settings Under Uncertainty
• When Should You Defend Your Classifier?
• Optimal training for adversarial games
• A fast saddle-point dynamical system approach to robust deep learning

Cites work

• A Further Generalization of the Kakutani Fixed Point Theorem, with Application to Nash Equilibrium Points
• Adversarial Risk via Optimal Transport and Optimal Couplings
• Game theory
• Game theory and machine learning for cyber security
• scientific article; zbMATH DE number 4141836 (Why is no real title available?)
• scientific article; zbMATH DE number 3228304 (Why is no real title available?)
• Multiagent Systems
• On the Stackelberg strategy in nonzero-sum games
• Why do deep convolutional networks generalize so poorly to small image transformations?

Cited in

(8)

• Optimal training for adversarial games
• Theoretical investigation of generalization bounds for adversarial learning of deep neural networks
• Dynamic Cheap Talk for Robust Adversarial Learning
• Adversarial models for priority‐based networks
• A Game Theoretic Perspective on Adversarial Machine Learning and Related Cybersecurity Applications
• A fast saddle-point dynamical system approach to robust deep learning
• When Should You Defend Your Classifier?
• Neural operators can play dynamic Stackelberg games

This page was built for publication: Achieving optimal adversarial accuracy for adversarial deep learning using Stackelberg games

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2080981)