arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2003.05012cs.LGcs.AIstat.ML
收藏

2019年MineRL样本高效强化学习竞赛回顾分析

Retrospective Analysis of the 2019 MineRL Competition on Sample Efficient Reinforcement Learning

Carnegie Mellon University（卡内基梅隆大学）；OpenAI；AIcrowd；Preferred Networks；DeepMind；Microsoft Research（微软研究院）

Stephanie Milani, Nicholay Topin, Brandon Houghton, William H. Guss, Sharada P. Mohanty, Keisuke Nakata, Oriol Vinyals, Noboru Sean Kuno

备注：
To appear in Proceedings of Machine Learning Research: NeurIPS 2019 Competition & Demonstration Track Postproceedings. 12 pages, 2 figures
阅读 PDF
arXiv 页面 ↗HTML 原文 ↗

摘要：

为了促进样本高效强化学习方向的研究，我们在第三十三届神经信息处理系统会议（NeurIPS 2019）上举办了MineRL竞赛。该竞赛的主要目标是促进使用人类演示和强化学习的算法的开发，以减少解决复杂，分层和稀疏环境所需的样本数量。我们描述了比赛，概述了主要挑战，比赛设计，以及我们为参与者提供的资源。我们提供了顶级解决方案的概述，每个解决方案都使用深度强化学习和/或模仿学习。我们还讨论了我们的组织决策对竞争和未来改进方向的影响。

英文摘要：

To facilitate research in the direction of sample efficient reinforcement learning, we held the MineRL Competition on Sample Efficient Reinforcement Learning Using Human Priors at the Thirty-third Conference on Neural Information Processing Systems (NeurIPS 2019). The primary goal of this competition was to promote the development of algorithms that use human demonstrations alongside reinforcement learning to reduce the number of samples needed to solve complex, hierarchical, and sparse environments. We describe the competition, outlining the primary challenge, the competition design, and the resources that we provided to the participants. We provide an overview of the top solutions, each of which use deep reinforcement learning and/or imitation learning. We also discuss the impact of our organizational decisions on the competition and future directions for improvement.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。

↑