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

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

Machine Learning Research: NeurIPS2019 Competition & Demonstration Track |

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.

MineRL Competition 2019

Starting June 1st, we are holding a competition on sample-efficient reinforcement learning using human priors. In our competition, participants develop a system to obtain a diamond in Minecraft using only four days of training time. See more at https://aka.ms/minerl