B2RL: An open-source Dataset for Building Batch Reinforcement Learning.
Hsin-Yu LiuXiaohan FuBharathan BalajiRajesh GuptaDezhi HongPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- open source
- function approximation
- reinforcement learning algorithms
- model free
- multi agent
- control problems
- markov decision processes
- rl algorithms
- state space
- temporal difference
- optimal policy
- open source software
- learning algorithm
- direct policy search
- batch mode
- case study
- reinforcement learning methods
- machine learning
- partially observable domains
- optimal control
- dynamic programming
- supervised learning
- temporal difference learning
- action selection
- continuous state
- policy gradient
- actor critic
- learning problems
- function approximators
- source code
- learned knowledge
- benchmark datasets
- autonomous learning
- approximate dynamic programming
- policy search
- continuous state and action spaces
- learning classifier systems