The Challenges of Exploration for Offline Reinforcement Learning.
Nathan LambertMarkus WulfmeierWilliam F. WhitneyArunkumar ByravanMichael BloeschVibhavari DasagiTim HertweckMartin A. RiedmillerPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- active exploration
- lessons learned
- exploration strategy
- model based reinforcement learning
- key issues
- action selection
- real world
- open issues
- markov decision processes
- application scenarios
- data mining
- state space
- real time
- active learning
- function approximation
- multi agent
- autonomous learning
- optimal policy
- evolutionary algorithm
- learning process
- learning algorithm
- stochastic approximation
- multi agent reinforcement learning
- genetic algorithm
- data sets