SEIHAI: A Sample-efficient Hierarchical AI for the MineRL Competition

2021/11/17 by Hangyu Mao, Mao, Hangyu, Chao Wang +15 · 2 citations
Computer Science · Engineering · Psychology · #Adversarial Robustness in Machine Learning #Artificial Intelligence (cs.AI) #Artificial intelligence #Competition (biology) #Computer science #Economics #FOS: Computer and information sciences #FOS: Electrical engineering #Imitation #Leverage (statistics) #Machine Learning (cs.LG) #Management #Multiagent Systems (cs.MA) #Physics #Psychology #Reinforcement Learning in Robotics #Reinforcement learning #Robot Manipulation and Learning #Robotics (cs.RO) #Sample (material) #Systems and Control (eess.SY) #Task (project management) #cs.AI #cs.LG #cs.MA #cs.RO #cs.SY #eess.SY #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2111.08857

published in arXiv (Cornell University) (Cornell University) · The winner solution of NeurIPS 2020 MineRL competition (https://www.aicrowd.com/challenges/neurips-2020-minerl-competition/leaderboards). The paper has been accepted by DAI 2021 (the third International Conference on Distributed Artificial Intelligence)

arxiv created 2021/11/17 · openalex publication_date 2021/11/17 · arxiv updated 2021/11/18 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/08

Abstract

The MineRL competition is designed for the development of reinforcement learning and imitation learning algorithms that can efficiently leverage human demonstrations to drastically reduce the number of environment interactions needed to solve the complex ObtainDiamond task with sparse rewards. To address the challenge, in this paper, we present SEIHAI, a Sample-efficient Hierarchical AI, that fully takes advantage of the human demonstrations and the task structure. Specifically, we split the task into several sequentially dependent subtasks, and train a suitable agent for each subtask using reinforcement learning and imitation learning. We further design a scheduler to select different agents for different subtasks automatically. SEIHAI takes the first place in the preliminary and final of the NeurIPS-2020 MineRL competition.

Citations

• The global k-means clustering algorithm
• Least squares quantization in PCM
• Stabilizing Off-Policy Q-Learning via Bootstrapping Error Reduction
• Deep Q-learning from Demonstrations

Cited by

• An Efficient Open World Environment for Multi-Agent Social Learning
• NGENT: Next-Generation AI Agents Must Integrate Multi-Domain Abilities to Achieve Artificial General Intelligence

Related

• The MineRL 2020 Competition on Sample Efficient Reinforcement Learning using Human Priors
• Retrospective Analysis of the 2019 MineRL Competition on Sample Efficient Reinforcement Learning
• The MineRL 2019 Competition on Sample Efficient Reinforcement Learning using Human Priors
• Retrospective on the 2021 BASALT Competition on Learning from Human Feedback
• Towards robust and domain agnostic reinforcement learning competitions