List your company
ResearchFormerly WizWand

Adversarial Detection on PGD perturbations (val)

99.5AUROCMMD-CC + SimCLRv2
16.8238.28559.7581.215Oct 4, 2022
Updated 8mo ago

Evaluation results

MethodLinks
MMD-CC + SimCLRv2n_samples=20, Backbone=ResNet-502022.1099.5
MMD + SimCLRv2n_samples=20, Backbone=ResNet-502022.1098.8
MMD-CC + SimCLRv2n_samples=10, Backbone=ResNet-502022.1096.6
MMD + SimCLRv2n_samples=10, Backbone=ResNet-502022.1086.6
MMD-CC + Supervisedn_samples=20, Backbone=ResNet-502022.1085.8
MMD-CC + SimCLRv2n_samples=5, Backbone=ResNet-502022.1084
MMD-CC + SimCLRv2n_samples=3, Backbone=ResNet-502022.1070.5
MMD-CC + Supervisedn_samples=10, Backbone=ResNet-502022.1070.5
MMD-CC + Supervisedn_samples=5, Backbone=ResNet-502022.1061.3
MMD + Supervisedn_samples=20, Backbone=ResNet-502022.1057.5
MMD-CC + Supervisedn_samples=3, Backbone=ResNet-502022.1057.4
MMD + SimCLRv2n_samples=5, Backbone=ResNet-502022.1053.8
MMD + SimCLRv2n_samples=3, Backbone=ResNet-502022.1035.2
MMD + Supervisedn_samples=10, Backbone=ResNet-502022.1033
MMD + Supervisedn_samples=5, Backbone=ResNet-502022.1022.5
MMD + Supervisedn_samples=3, Backbone=ResNet-502022.1020