Adversarial Attack Metrics Profile on CIFAR-10
88.3AccuracyZOO
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ZOOPrecision=DNN (32-bit floating-point), Attack Category=Gray-Box2026.06 | 88.3 | 24.7 | 63.6 | 0.05 | |
| Square-L2Precision=DNN (32-bit floating-point), Attack Category=Gray-Box2026.06 | 88.3 | 47.3 | 41 | 0.22 | |
| Square-L∞Precision=DNN (32-bit floating-point), Attack Category=Gray-Box2026.06 | 88.3 | 6.1 | 82.2 | 0.01 | |
| BoundaryPrecision=DNN (32-bit floating-point), Attack Category=Black-Box2026.06 | 88.3 | 7.7 | 80.6 | 0.01 | |
| GeoDAPrecision=DNN (32-bit floating-point), Attack Category=Black-Box2026.06 | 88.3 | 11 | 77.3 | 0.03 | |
| ZOOPrecision=QNN (8-bit integer), Attack Category=Gray-Box2026.06 | 87.5 | 87.5 | 0 | 0 | |
| Square-L2Precision=QNN (8-bit integer), Attack Category=Gray-Box2026.06 | 87.5 | 67.8 | 19.7 | 0.22 | |
| Square-L∞Precision=QNN (8-bit integer), Attack Category=Gray-Box2026.06 | 87.5 | 82.3 | 5.2 | 0.01 | |
| BoundaryPrecision=QNN (8-bit integer), Attack Category=Black-Box2026.06 | 87.5 | 6.5 | 81 | 0.08 | |
| GeoDAPrecision=QNN (8-bit integer), Attack Category=Black-Box2026.06 | 87.5 | 13 | 74.5 | 0.08 | |
| PGDPrecision=QNN (8-bit integer), Attack Category=Extraction → White-Box2026.06 | 85.3 | 38 | 47.3 | 0.03 | |
| FGSMPrecision=QNN (8-bit integer), Attack Category=Extraction → White-Box2026.06 | 85.3 | 50 | 35.3 | 0.03 | |
| PGDPrecision=DNN (32-bit floating-point), Attack Category=Extraction → White-Box2026.06 | 84.8 | 40.2 | 44.6 | 0.03 | |
| FGSMPrecision=DNN (32-bit floating-point), Attack Category=Extraction → White-Box2026.06 | 84.8 | 51.1 | 33.7 | 0.03 |