Text Classification on SST-2 (CA, ASR, FRR, FAR)
95.06CASTRIP
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| STRIPAttack=BadNets, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 95.06 | 94.38 | 7.56 | 87.44 | |
| ONIONAttack=BadNets, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 95.06 | 94.38 | 2.78 | 9.28 | |
| RAPAttack=BadNets, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 95.06 | 94.38 | 3.11 | 64.28 | |
| MDPAttack=BadNets, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 95.06 | 94.38 | 5.33 | 1.77 | |
| STRIPAttack=AddSent, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 94.45 | 100 | 2.75 | 72.56 | |
| ONIONAttack=AddSent, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 94.45 | 100 | 7.06 | 26.72 | |
| RAPAttack=AddSent, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 94.45 | 100 | 5.61 | 37.5 | |
| MDPAttack=AddSent, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 94.45 | 100 | 4.45 | 3.53 | |
| STRIPAttack=EP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 93.63 | 95.95 | 1.72 | 72.06 | |
| ONIONAttack=EP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 93.63 | 95.95 | 5.28 | 12.89 | |
| RAPAttack=EP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 93.63 | 95.95 | 2.72 | 58.11 | |
| MDPAttack=EP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 93.63 | 95.95 | 5.05 | 0.73 | |
| STRIPAttack=LWP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 93.41 | 95.53 | 5.96 | 89.39 | |
| ONIONAttack=LWP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 93.41 | 95.53 | 8.28 | 7.39 | |
| RAPAttack=LWP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 93.41 | 95.53 | 0.83 | 43.77 | |
| MDPAttack=LWP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 93.41 | 95.53 | 5.27 | 4.78 | |
| STRIPAttack=SOS, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 91.65 | 92.41 | 2.98 | 87.56 | |
| ONIONAttack=SOS, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 91.65 | 92.41 | 4.06 | 32.56 | |
| RAPAttack=SOS, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 91.65 | 92.41 | 1.89 | 51.28 | |
| MDPAttack=SOS, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 91.65 | 92.41 | 0 | 0 |