Adaptively Calibrated Critic Estimates for Deep Reinforcement Learning.
Nicolai DorkaJoschka BoedeckerWolfram BurgardPublished in: CoRR (2021)
Keyphrases
- reinforcement learning
- function approximation
- temporal difference
- reinforcement learning algorithms
- actor critic
- policy gradient
- state space
- model free
- learning algorithm
- dynamic programming
- learning process
- temporal difference learning
- natural actor critic
- importance sampling
- markov decision processes
- transition model
- policy search
- confidence intervals
- markov decision process
- optimal control
- real valued
- approximate dynamic programming
- joint estimation
- robotic control
- multi view