Reinforcement Learning on Maze 17^10 structured discrete
9.61Mean ScoreDGRL (SDN, DBU)
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DGRL (SDN, DBU)Num Seeds=102026.02 | 9.61 | 9.63 | 0.04 | 0 | 19,548 | |
| SDN (EU), DBUNum Seeds=102026.02 | 9.57 | 9.6 | 0.08 | 0 | 19,494 | |
| SDN (MH), DBUNum Seeds=102026.02 | 9.56 | 9.61 | 0.14 | 0 | 19,505 | |
| DNC (SA), DBUNum Seeds=102026.02 | 9.16 | 9.22 | 0.16 | -4 | 18,730 | |
| DNC (SA), A2CNum Seeds=102026.02 | 8.79 | 9.69 | 1.33 | 0 | 19,668 | |
| SDN, A2CNum Seeds=102026.02 | 8.65 | 8.73 | 0.42 | -9 | 17,740 | |
| SDN, DDPGNum Seeds=102026.02 | 7.88 | 8.79 | 2.33 | -9 | 17,863 | |
| DNC (greedy), A2CNum Seeds=102026.02 | 6.75 | 7.28 | 2.72 | -24 | 14,804 | |
| Cacla, A2CNum Seeds=102026.02 | 1.15 | -0.05 | 2.01 | -100 | 0 |