Offline Reinforcement Learning on Maze2D umaze
141Normalized ReturnGPC-SAC
Evaluation Results
| Method | Links | |
|---|---|---|
| GPC-SAC2026.02 | 141 | |
| OURS2026.02 | 136.9 | |
| PBRL2026.02 | 86.7 | |
| Real DatasetOffline RL Algorithm=TD3PLUSBC, Privacy Budget (epsilon)=no-DP2025.12 | 73.6 | |
| Real DatasetOffline RL Algorithm=IQL, Privacy Budget (epsilon)=no-DP2025.12 | 71 | |
| Real DatasetOffline RL Algorithm=EDAC, Privacy Budget (epsilon)=no-DP2025.12 | 70.8 | |
| PrivORL-nOffline RL Algorithm=IQL, Privacy Budget (epsilon)=102025.12 | 70.3 | |
| PrivORL-nOffline RL Algorithm=EDAC, Privacy Budget (epsilon)=102025.12 | 69.1 | |
| PrivORL-nOffline RL Algorithm=EDAC, Privacy Budget (epsilon)=12025.12 | 63.2 | |
| PrivORL-nOffline RL Algorithm=TD3PLUSBC, Privacy Budget (epsilon)=102025.12 | 60.3 | |
| PrivORL-nOffline RL Algorithm=IQL, Privacy Budget (epsilon)=12025.12 | 60.1 | |
| CQL2026.02 | 56.3 | |
| IQL2026.02 | 46.9 | |
| PGMOffline RL Algorithm=IQL, Privacy Budget (epsilon)=102025.12 | 41.6 | |
| PGMOffline RL Algorithm=EDAC, Privacy Budget (epsilon)=102025.12 | 32.8 | |
| PrivORL-nOffline RL Algorithm=TD3PLUSBC, Privacy Budget (epsilon)=12025.12 | 30.6 | |
| PGMOffline RL Algorithm=IQL, Privacy Budget (epsilon)=12025.12 | 30.1 | |
| PrePATE-GANOffline RL Algorithm=TD3PLUSBC, Privacy Budget (epsilon)=102025.12 | 26.4 | |
| PrePATE-GANOffline RL Algorithm=IQL, Privacy Budget (epsilon)=102025.12 | 20.2 | |
| PrePATE-GANOffline RL Algorithm=EDAC, Privacy Budget (epsilon)=102025.12 | 18.4 | |
| PGMOffline RL Algorithm=EDAC, Privacy Budget (epsilon)=12025.12 | 17.2 | |
| PrePATE-GANOffline RL Algorithm=TD3PLUSBC, Privacy Budget (epsilon)=12025.12 | 15.3 | |
| PrePATE-GANOffline RL Algorithm=IQL, Privacy Budget (epsilon)=12025.12 | 12.4 | |
| PrePATE-GANOffline RL Algorithm=EDAC, Privacy Budget (epsilon)=12025.12 | 12.1 | |
| PATE-GANOffline RL Algorithm=TD3PLUSBC, Privacy Budget (epsilon)=102025.12 | 9.1 | |
| PrivSynOffline RL Algorithm=TD3PLUSBC, Privacy Budget (epsilon)=102025.12 | 5.7 | |
| PrivSynOffline RL Algorithm=TD3PLUSBC, Privacy Budget (epsilon)=12025.12 | 3 | |
| PrivSynOffline RL Algorithm=IQL, Privacy Budget (epsilon)=102025.12 | 2.9 | |
| PATE-GANOffline RL Algorithm=TD3PLUSBC, Privacy Budget (epsilon)=12025.12 | 1.5 | |
| PrivSynOffline RL Algorithm=EDAC, Privacy Budget (epsilon)=102025.12 | 1.3 | |
| PrivSynOffline RL Algorithm=EDAC, Privacy Budget (epsilon)=12025.12 | 0.1 | |
| PrivSynOffline RL Algorithm=IQL, Privacy Budget (epsilon)=12025.12 | 0 | |
| PATE-GANOffline RL Algorithm=EDAC, Privacy Budget (epsilon)=12025.12 | -8.2 | |
| PGMOffline RL Algorithm=TD3PLUSBC, Privacy Budget (epsilon)=102025.12 | -8.5 | |
| PATE-GANOffline RL Algorithm=IQL, Privacy Budget (epsilon)=12025.12 | -10.5 | |
| PGMOffline RL Algorithm=TD3PLUSBC, Privacy Budget (epsilon)=12025.12 | -12.4 | |
| PATE-GANOffline RL Algorithm=IQL, Privacy Budget (epsilon)=102025.12 | -14.4 | |
| PATE-GANOffline RL Algorithm=EDAC, Privacy Budget (epsilon)=102025.12 | -16.8 |