Parrot: Data-Driven Behavioral Priors for Reinforcement Learning.
Avi SinghHuihan LiuGaoyue ZhouAlbert YuNicholas RhinehartSergey LevinePublished in: ICLR (2021)
Keyphrases
- data driven
- reinforcement learning
- function approximation
- selective perception
- temporal difference
- optimal policy
- learning algorithm
- state space
- bayesian framework
- decision making
- reinforcement learning algorithms
- agent behavior
- temporal difference learning
- markov decision process
- robot control
- reinforcement learning methods
- partially observable
- stochastic approximation
- markov decision processes
- maximum a posteriori
- real time
- dynamic programming
- prior knowledge
- learning process
- case study
- information systems
- machine learning
- neural network