NeoRL: A Near Real-World Benchmark for Offline Reinforcement Learning.
Rongjun QinXingyuan ZhangSongyi GaoXiong-Hui ChenZewen LiWeinan ZhangYang YuPublished in: NeurIPS (2022)
Keyphrases
- real world
- reinforcement learning
- wide range
- data mining
- real time
- learning algorithm
- data sets
- learning process
- machine learning
- real life
- case study
- database
- markov decision processes
- function approximation
- multiscale
- multi agent
- support vector
- database systems
- state space
- dynamic environments
- optimal policy
- model free
- temporal difference
- real robot
- temporal difference learning