Text Classification on MR (Error Rate Metrics)
89.85CASTRIP
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| STRIPAttack=SOS, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.85 | 97.3 | 5.2 | 75.9 | |
| ONIONAttack=SOS, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.85 | 97.3 | 0.9 | 64.1 | |
| RAPAttack=SOS, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.85 | 97.3 | 15.2 | 58.85 | |
| MDPAttack=SOS, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.85 | 97.3 | 4.85 | 3.4 | |
| STRIPAttack=BadNets, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.8 | 98.3 | 11.7 | 72.3 | |
| ONIONAttack=BadNets, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.8 | 98.3 | 4.8 | 15.6 | |
| RAPAttack=BadNets, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.8 | 98.3 | 2.75 | 25.35 | |
| MDPAttack=BadNets, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.8 | 98.3 | 5.1 | 5.6 | |
| STRIPAttack=LWP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.65 | 96.9 | 9.35 | 82.7 | |
| ONIONAttack=LWP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.65 | 96.9 | 1.6 | 17.45 | |
| RAPAttack=LWP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.65 | 96.9 | 1.7 | 52.55 | |
| MDPAttack=LWP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.65 | 96.9 | 5.25 | 3.6 | |
| STRIPAttack=AddSent, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.6 | 97.5 | 16.2 | 60 | |
| ONIONAttack=AddSent, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.6 | 97.5 | 4.65 | 37.25 | |
| RAPAttack=AddSent, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.6 | 97.5 | 9.35 | 39.7 | |
| MDPAttack=AddSent, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.6 | 97.5 | 5.05 | 10.9 | |
| STRIPAttack=EP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.4 | 96.6 | 2.2 | 88.9 | |
| ONIONAttack=EP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.4 | 96.6 | 15.35 | 12.6 | |
| RAPAttack=EP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.4 | 96.6 | 6.45 | 70.6 | |
| MDPAttack=EP, Learning Strategy=prompt-based fine-tuning, FRR threshold=5%2023.09 | 89.4 | 96.6 | 4.7 | 3 |