Reinforcement Learning on Maze 5^4 structured discrete
9.74Mean ScoreDNC (SA), A2C
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DNC (SA), A2CNum Seeds=102026.02 | 9.74 | 9.75 | 0.03 | 0 | 14,507 | |
| DGRL (SDN, DBU)Num Seeds=102026.02 | 9.67 | 9.67 | 0.01 | 0 | 14,390 | |
| DNC (greedy), A2CNum Seeds=102026.02 | 9.35 | 9.36 | 0.17 | -3 | 13,927 | |
| Wolpertinger, DBUNum Seeds=102026.02 | 8.31 | 8.63 | 1.13 | -11 | 12,822 | |
| Wolpertinger, DDPGNum Seeds=102026.02 | 8.24 | 8.91 | 1.29 | -8 | 13,248 | |
| Cacla, A2CNum Seeds=102026.02 | 1.36 | 0.07 | 2.31 | -99 | 0 | |
| VAC, A2CNum Seeds=102026.02 | 0.99 | 0.98 | 0.64 | -89 | 1,374 |