List your company
ResearchFormerly WizWand

Membership Inference Attack on CIFAR-100 Low forget set

71.633MIA Correctness RateSCRUB
-2.7613216.5525935.866555.18041Jun 16, 2025
Updated 8mo ago

Evaluation results

MethodLinks
SCRUBOptimization Algorithm (A)=ASAM 0.1, Unlearning enhancement=None2025.0671.633
SCRUBOptimization Algorithm (A)=SAM 0.1, Unlearning enhancement=None2025.0656.3
SCRUBOptimization Algorithm (A)=SGD, Unlearning enhancement=None2025.0632.6
RLOptimization Algorithm (A)=ASAM 1.0, Unlearning enhancement=None2025.0615.8
RLOptimization Algorithm (A)=ASAM 0.1, Unlearning enhancement=None2025.0613.5
RLOptimization Algorithm (A)=ASAM 0.1, Unlearning enhancement=ASAM 1.02025.0612.7
SalUnOptimization Algorithm (A)=ASAM 0.1, Unlearning enhancement=None2025.0611.167
RLOptimization Algorithm (A)=SGD, Unlearning enhancement=None2025.0610.767
RLOptimization Algorithm (A)=ASAM 1.0, Unlearning enhancement=ASAM 1.02025.0610.667
SCRUBOptimization Algorithm (A)=ASAM 1.0, Unlearning enhancement=None2025.069.833
RLOptimization Algorithm (A)=SGD, Unlearning enhancement=ASAM 1.02025.069.467
SalUnOptimization Algorithm (A)=ASAM 1.0, Unlearning enhancement=None2025.068.8
L1-SparseOptimization Algorithm (A)=SGD, Unlearning enhancement=None2025.068.6
RLOptimization Algorithm (A)=SAM 0.1, Unlearning enhancement=None2025.068.267
SalUnOptimization Algorithm (A)=SGD, Unlearning enhancement=None2025.067.333
SalUnOptimization Algorithm (A)=SAM 0.1, Unlearning enhancement=None2025.066.867
SalUnOptimization Algorithm (A)=ASAM 0.1, Unlearning enhancement=ASAM 1.02025.066.633
RLOptimization Algorithm (A)=SAM 0.1, Unlearning enhancement=ASAM 1.02025.065.833
L1-SparseOptimization Algorithm (A)=ASAM 1.0, Unlearning enhancement=None2025.065.8
L1-SparseOptimization Algorithm (A)=SGD, Unlearning enhancement=ASAM 1.02025.065.5
L1-SparseOptimization Algorithm (A)=SAM 0.1, Unlearning enhancement=None2025.065.033
SalUnOptimization Algorithm (A)=SGD, Unlearning enhancement=ASAM 1.02025.064.5
SalUnOptimization Algorithm (A)=ASAM 1.0, Unlearning enhancement=ASAM 1.02025.064.2
SalUnOptimization Algorithm (A)=SAM 0.1, Unlearning enhancement=ASAM 1.02025.064.067
L1-SparseOptimization Algorithm (A)=ASAM 0.1, Unlearning enhancement=None2025.064
L1-SparseOptimization Algorithm (A)=ASAM 1.0, Unlearning enhancement=ASAM 1.02025.063.9
L1-SparseOptimization Algorithm (A)=SAM 0.1, Unlearning enhancement=ASAM 1.02025.063.8
L1-SparseOptimization Algorithm (A)=ASAM 0.1, Unlearning enhancement=ASAM 1.02025.063.6
NegGradOptimization Algorithm (A)=SAM 0.1, Unlearning enhancement=None2025.063.433
NegGradOptimization Algorithm (A)=SGD, Unlearning enhancement=None2025.062.167
NegGradOptimization Algorithm (A)=ASAM 1.0, Unlearning enhancement=None2025.061.8
NegGradOptimization Algorithm (A)=ASAM 0.1, Unlearning enhancement=None2025.061.733
NegGradOptimization Algorithm (A)=SGD, Unlearning enhancement=ASAM 1.02025.060.733
NegGradOptimization Algorithm (A)=ASAM 0.1, Unlearning enhancement=ASAM 1.02025.060.7
NegGradOptimization Algorithm (A)=ASAM 1.0, Unlearning enhancement=ASAM 1.02025.060.533
NegGradOptimization Algorithm (A)=SAM 0.1, Unlearning enhancement=ASAM 1.02025.060.366
SCRUBOptimization Algorithm (A)=ASAM 1.0, Unlearning enhancement=ASAM 1.02025.060.2
SCRUBOptimization Algorithm (A)=SAM 0.1, Unlearning enhancement=ASAM 1.02025.060.2
SCRUBOptimization Algorithm (A)=ASAM 0.1, Unlearning enhancement=ASAM 1.02025.060.167
SCRUBOptimization Algorithm (A)=SGD, Unlearning enhancement=ASAM 1.02025.060.1