Atari Game Reinforcement Learning on Atari Seaquest Unexpected Environment (evaluation)
75.6RewardPPO + Repaired Shield
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PPO + Repaired ShieldShield type=Repaired2025.11 | 75.6 | 100 | 0.05 | |
| PPO + Adaptive ShieldShield type=Adaptive2025.11 | 58.8 | 100 | 0.06 | |
| PPO + Naive ShieldShield type=Naive2025.11 | 0.2 | 1 | 0.18 | |
| PPOShield type=None2025.11 | 0 | 0 | — | |
| PPO + Static ShieldShield type=Static2025.11 | 0 | 0 | 0.08 |