论文标题
Atari-5：将街机学习环境提炼到五场比赛
Atari-5: Distilling the Arcade Learning Environment down to Five Games
论文作者
论文摘要
街机学习环境（ALE）已成为评估增强学习算法的性能的重要基准。但是，在整个57游戏数据集中生成结果的计算成本限制了ALE的使用，并使许多结果的可重复性变得不可行。我们以一种原则性方法的形式提出了一种新的解决方案，以选择基准套件中环境的小但代表性的子集的形式。我们应用了我们的方法来识别五个ALE游戏的子集，称为Atari-5，该游戏的中位数得分估计值在其真实值的10％以内。将子集扩展到10游戏可恢复57场比赛中所有游戏的日志评分方差的80％。由于啤酒中的许多游戏之间的高度相关性，我们可以显示这种压缩水平。
The Arcade Learning Environment (ALE) has become an essential benchmark for assessing the performance of reinforcement learning algorithms. However, the computational cost of generating results on the entire 57-game dataset limits ALE's use and makes the reproducibility of many results infeasible. We propose a novel solution to this problem in the form of a principled methodology for selecting small but representative subsets of environments within a benchmark suite. We applied our method to identify a subset of five ALE games, called Atari-5, which produces 57-game median score estimates within 10% of their true values. Extending the subset to 10-games recovers 80% of the variance for log-scores for all games within the 57-game set. We show this level of compression is possible due to a high degree of correlation between many of the games in ALE.
