MiniHack the Planet: A Sandbox for Open-Ended Reinforcement Learning Research.
Mikayel SamvelyanRobert KirkVitaly KurinJack Parker-HolderMinqi JiangEric HambroFabio PetroniHeinrich KüttlerEdward GrefenstetteTim RocktäschelPublished in: NeurIPS Datasets and Benchmarks (2021)
Keyphrases
- open ended
- reinforcement learning
- function approximation
- learning outcomes
- model free
- optimal policy
- markov decision processes
- inquiry learning
- reinforcement learning algorithms
- machine learning
- learning algorithm
- state space
- dynamic programming
- temporal difference
- multiple choice
- learning processes
- learning process
- multi agent