Image Classification on CIFAR-10 (CA and ASR Evaluation)
93.62AccuracyNo defense
Evaluation results
| Method | Links | ||
|---|---|---|---|
| No defenseBackbone=MobileNetV2, Quantization=8-bit, Attack Type=Exploiting LLM2026.06 | 93.62 | 97.31 | |
| No defenseBackbone=MobileNetV2, Quantization=4-bit, Attack Type=Exploiting LLM2026.06 | 92.87 | 97.34 | |
| Gaussian NoiseBackbone=MobileNetV2, Quantization=8-bit, Attack Type=Exploiting LLM2026.06 | 92.84 | 5.78 | |
| LACBackbone=MobileNetV2, Quantization=8-bit, Attack Type=Exploiting LLM2026.06 | 92.32 | 0.53 | |
| EFRAPBackbone=MobileNetV2, Quantization=8-bit, Attack Type=Exploiting LLM2026.06 | 92.29 | 0.5 | |
| QVec (0.98)Backbone=MobileNetV2, Quantization=8-bit, Attack Type=Exploiting LLM2026.06 | 91.06 | 0.21 | |
| QVec (0.98)Backbone=MobileNetV2, Quantization=4-bit, Attack Type=Exploiting LLM2026.06 | 90.94 | 18.2 | |
| QVec (0.95)Backbone=MobileNetV2, Quantization=8-bit, Attack Type=Exploiting LLM2026.06 | 90.55 | 0.14 | |
| QVec (0.98)Backbone=MobileNetV2, Quantization=8-bit, Attack Type=Qu-Anti-zation2026.06 | 90.26 | 5.21 | |
| LACBackbone=MobileNetV2, Quantization=4-bit, Attack Type=Exploiting LLM2026.06 | 89.03 | 5.39 | |
| QVec (0.95)Backbone=MobileNetV2, Quantization=8-bit, Attack Type=Qu-Anti-zation2026.06 | 89.01 | 5.18 | |
| EFRAPBackbone=MobileNetV2, Quantization=4-bit, Attack Type=Exploiting LLM2026.06 | 88.83 | 5.42 | |
| EFRAPBackbone=MobileNetV2, Quantization=8-bit, Attack Type=Qu-Anti-zation2026.06 | 88.28 | 1.32 | |
| LACBackbone=MobileNetV2, Quantization=8-bit, Attack Type=Qu-Anti-zation2026.06 | 88.2 | 1.32 | |
| QVec (0.95)Backbone=MobileNetV2, Quantization=4-bit, Attack Type=Exploiting LLM2026.06 | 87.53 | 9.3 | |
| No defenseBackbone=MobileNetV2, Quantization=8-bit, Attack Type=Qu-Anti-zation2026.06 | 86.44 | 96.68 | |
| LACBackbone=MobileNetV2, Quantization=4-bit, Attack Type=Qu-Anti-zation2026.06 | 85.45 | 1.33 | |
| EFRAPBackbone=MobileNetV2, Quantization=4-bit, Attack Type=Qu-Anti-zation2026.06 | 85.09 | 1.47 | |
| QVec (0.98)Backbone=MobileNetV2, Quantization=4-bit, Attack Type=Qu-Anti-zation2026.06 | 82.17 | 14.3 | |
| QVec (0.95)Backbone=MobileNetV2, Quantization=4-bit, Attack Type=Qu-Anti-zation2026.06 | 79.6 | 11.07 | |
| Gaussian NoiseBackbone=MobileNetV2, Quantization=4-bit, Attack Type=Qu-Anti-zation2026.06 | 71.1 | 93.31 | |
| Gaussian NoiseBackbone=MobileNetV2, Quantization=4-bit, Attack Type=Exploiting LLM2026.06 | 69.81 | 32.79 | |
| No defenseBackbone=MobileNetV2, Quantization=4-bit, Attack Type=Qu-Anti-zation2026.06 | 67.92 | 97.88 | |
| Gaussian NoiseBackbone=MobileNetV2, Quantization=8-bit, Attack Type=Qu-Anti-zation2026.06 | 66.22 | 34.6 |