Model Extraction on CIFAR-10 (test)
92.54AccuracyNo Def.
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| No Def.Budget=Victim Acc.2026.05 | 92.54 | — | — | — | — | |
| PPBudget=Victim Acc.2026.05 | 92.54 | — | — | — | — | |
| CIPBudget=Victim Acc.2026.05 | 92.54 | — | — | — | — | |
| LymphNodeBudget=Victim Acc.2026.05 | 92.54 | — | — | — | — | |
| No Def.Budget=50,0002026.05 | 85.24 | — | — | — | — | |
| No Def.Budget=10,0002026.05 | 66.16 | — | — | — | — | |
| No Def.Budget=5,0002026.05 | 56.25 | — | — | — | — | |
| CIPBudget=50,0002026.05 | 48.25 | — | — | — | — | |
| PPBudget=50,0002026.05 | 45.66 | — | — | — | — | |
| No Def.Budget=1,0002026.05 | 42.88 | — | — | — | — | |
| CIPBudget=10,0002026.05 | 42.81 | — | — | — | — | |
| PPBudget=10,0002026.05 | 38.24 | — | — | — | — | |
| CIPBudget=5,0002026.05 | 35.17 | — | — | — | — | |
| PPBudget=5,0002026.05 | 32.68 | — | — | — | — | |
| CIPBudget=1,0002026.05 | 28.44 | — | — | — | — | |
| PPBudget=1,0002026.05 | 25.26 | — | — | — | — | |
| LymphNodeBudget=50,0002026.05 | 15.28 | — | — | — | — | |
| LymphNodeBudget=5,0002026.05 | 14.76 | — | — | — | — | |
| LymphNodeBudget=10,0002026.05 | 14.68 | — | — | — | — | |
| LymphNodeBudget=1,0002026.05 | 13.71 | — | — | — | — | |
| CCSH+25Architecture=MLP-b 3072-256-256-256-64-10, Parameters=935,370, Data Precision=64-bit2024.11 | — | — | 2 | — | — | |
| CDG+25Architecture=MLP-a 1024-2-2-1, Parameters=2,059, Data Precision=64-bit2024.11 | — | 2 | — | 2 | 100 | |
| Divide-and-Conquer Side-Channel AttackArchitecture=MLP-a 1024-2-2-1, Parameters=2,059, Data Precision=64-bit2024.11 | — | 2 | 2 | 2 | 99.5 | |
| Divide-and-Conquer Side-Channel AttackArchitecture=MLP-b 3072-256-256-256-64-10, Parameters=935,370, Data Precision=64-bit2024.11 | — | 2 | 2 | 2 | 97.2 | |
| Divide-and-Conquer Side-Channel AttackArchitecture=MLP-c 3072-512-256-64-10, Parameters=1,721,802, Data Precision=32-bit2024.11 | — | 2 | 2 | 2 | 93.2 | |
| Divide-and-Conquer Side-Channel AttackArchitecture=Truncated MobileNetv1, Parameters=5,234, Data Precision=32-bit2024.11 | — | 2 | 2 | 2 | 88.4 |