SOTAVerified

Adversarial Attack

An Adversarial Attack is a technique to find a perturbation that changes the prediction of a machine learning model. The perturbation can be very small and imperceptible to human eyes.

Source: Recurrent Attention Model with Log-Polar Mapping is Robust against Adversarial Attacks

Papers

Recently AddedMost HypedMost ActiveNeeds VerificationMost Verified

Showing 1671–1680 of 1808 papers

TitleStatusHype
Delving into Data: Effectively Substitute Training for Black-box Attack—0
SNEAK: Synonymous Sentences-Aware Adversarial Attack on Natural Language Video Localization—0
Democratic Training Against Universal Adversarial Perturbations—0
Derivation of Information-Theoretically Optimal Adversarial Attacks with Applications to Robust Machine Learning—0
Design of secure and robust cognitive system for malware detection—0
Snowball Adversarial Attack on Traffic Sign Classification—0
Detecting Adversarial Directions in Deep Reinforcement Learning to Make Robust Decisions—0
Universal Attacks on Equivariant Networks—0
Solving Non-Convex Non-Differentiable Min-Max Games using Proximal Gradient Method—0
Adversarial Attack on Hierarchical Graph Pooling Neural Networks—0
Show:102550
← PrevPage 168 of 181Next →
All datasetsCIFAR-10CIFAR-100

Benchmark Results

#ModelMetricClaimedVerifiedStatus
1Xu et al.Attack: PGD2078.68—Unverified
23-ensemble of multi-resolution self-ensemblesAttack: AutoAttack78.13—Unverified
3TRADES-ANCRA/ResNet18Attack: AutoAttack59.7—Unverified
4AdvTraining [madry2018]Attack: PGD2048.44—Unverified
5TRADES [zhang2019b]Attack: PGD2045.9—Unverified
6XU-NetRobust Accuracy1—Unverified
#ModelMetricClaimedVerifiedStatus
13-ensemble of multi-resolution self-ensemblesAttack: AutoAttack51.28—Unverified
2multi-resolution self-ensemblesAttack: AutoAttack47.85—Unverified