Reinforcement Learning on Space Invaders Atari 2600 (test)
3,690Average Total RewardHuman
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Humanagent=expert human player2013.12 | 3,690 | — | |
| DQNpolicy=epsilon-greedy, epsilon=0.052013.12 | 581 | — | |
| Sarsapolicy=epsilon-greedy, epsilon=0.052013.12 | 271 | — | |
| Contingencypolicy=epsilon-greedy, epsilon=0.052013.12 | 268 | — | |
| Randompolicy=random2013.12 | 179 | — | |
| DQN Bestpolicy=epsilon-greedy, epsilon=0.05, evaluation=best episode2013.12 | — | 1,075 | |
| HNeat Bestpolicy=deterministic2013.12 | — | 1,720 | |
| HNeat Pixelpolicy=deterministic, input=8 color channel representation2013.12 | — | 1,145 |