Adversarial Attack on Breakout Black-box discrete (test)
130.26Cumulative RewardFGSM
Evaluation results
| Method | Links | |
|---|---|---|
| FGSMTarget Policy=PPO2026.02 | 130.26 | |
| FGSMTarget Policy=TRPO2026.02 | 128.24 | |
| PA-ADTarget Policy=A3C2026.02 | 123.63 | |
| FGSMTarget Policy=A3C2026.02 | 120.42 | |
| FGSMTarget Policy=DQN2026.02 | 118.32 | |
| PA-ADTarget Policy=TRPO2026.02 | 113.74 | |
| PIATarget Policy=PPO2026.02 | 100.47 | |
| PA-ADTarget Policy=PPO2026.02 | 99.87 | |
| PA-ADTarget Policy=DQN2026.02 | 98.48 | |
| SkipTarget Policy=PPO2026.02 | 95.42 | |
| SkipTarget Policy=TRPO2026.02 | 92.34 | |
| SkipTarget Policy=A3C2026.02 | 90.34 | |
| TSGETarget Policy=PPO2026.02 | 89.36 | |
| SkipTarget Policy=DQN2026.02 | 88.46 | |
| S-TTarget Policy=TRPO2026.02 | 70.24 | |
| S-TTarget Policy=PPO2026.02 | 69.21 | |
| S-TTarget Policy=DQN2026.02 | 68.32 | |
| S-TTarget Policy=A3C2026.02 | 68.21 | |
| PIATarget Policy=TRPO2026.02 | 63.47 | |
| EDGETarget Policy=PPO2026.02 | 60.84 | |
| EDGETarget Policy=A3C2026.02 | 60.36 | |
| EDGETarget Policy=TRPO2026.02 | 59.86 | |
| EDGETarget Policy=DQN2026.02 | 56.88 | |
| TSGETarget Policy=A3C2026.02 | 53.46 | |
| PIATarget Policy=DQN2026.02 | 53.45 | |
| TSGETarget Policy=TRPO2026.02 | 52.61 | |
| PIATarget Policy=A3C2026.02 | 48.72 | |
| AdvRL-GANTarget Policy=PPO2026.02 | 48.23 | |
| AdvRL-GANTarget Policy=TRPO2026.02 | 46.68 | |
| AdvRL-GANTarget Policy=A3C2026.02 | 45.28 | |
| TSGETarget Policy=DQN2026.02 | 43.73 | |
| AdvRL-GANTarget Policy=DQN2026.02 | 40.6 | |
| AATTarget Policy=DQN2026.02 | 28.32 | |
| AATTarget Policy=TRPO2026.02 | 22.22 | |
| AATTarget Policy=A3C2026.02 | 20.36 | |
| AATTarget Policy=PPO2026.02 | 20.21 |