Combining Model-Based and Model-Free Updates for Trajectory-Centric Reinforcement Learning.
Yevgen ChebotarKarol HausmanMarvin ZhangGaurav S. SukhatmeStefan SchaalSergey LevinePublished in: ICML (2017)
Keyphrases
- model free
- reinforcement learning
- reinforcement learning algorithms
- function approximation
- temporal difference
- reinforcement learning methods
- policy iteration
- policy evaluation
- state space
- learning algorithm
- rl algorithms
- transfer learning
- neural network
- average reward
- optimal policy
- fixed point
- learning process
- impedance control