B2RL: an open-source dataset for building batch reinforcement learning.
Hsin-Yu LiuXiaohan FuBharathan BalajiRajesh GuptaDezhi HongPublished in: BuildSys@SenSys (2022)
Keyphrases
- reinforcement learning
- open source
- open source software
- learning algorithm
- batch mode
- reinforcement learning algorithms
- model free
- function approximation
- state space
- rl algorithms
- markov decision processes
- multi agent
- machine learning
- partially observable domains
- temporal difference
- optimal control
- control problems
- continuous state
- optimal policy
- benchmark datasets
- dynamic programming
- approximate dynamic programming
- policy search
- batch processing
- direct policy search
- markov decision process
- complex domains
- partially observable
- case study
- learning problems
- transfer learning
- source code
- feature set
- policy iteration
- learning agents
- reinforcement learning methods
- learning process