Conservative Q-Learning for Offline Reinforcement Learning.
Aviral KumarAurick ZhouGeorge TuckerSergey LevinePublished in: CoRR (2020)
Keyphrases
- reinforcement learning
- function approximation
- reinforcement learning algorithms
- model free
- state space
- optimal policy
- state action space
- temporal difference learning
- temporal difference
- multi agent
- learning algorithm
- stochastic approximation
- multi agent reinforcement learning
- reinforcement learning methods
- markov decision processes
- continuous state
- machine learning
- eligibility traces
- policy iteration
- learning capabilities
- relational reinforcement learning
- function approximators
- optimal control
- supervised learning
- hierarchical reinforcement learning
- dynamic programming
- real time
- policy search
- reward shaping
- sequential decision problems
- exploration strategy
- rl algorithms
- learning agent
- control problems
- partially observable
- learning problems
- cooperative