Eigensubspace of Temporal-Difference Dynamics and How It Improves Value Approximation in Reinforcement Learning.
Qiang HeTianyi ZhouMeng FangSetareh MaghsudiPublished in: CoRR (2023)
Keyphrases
- temporal difference
- reinforcement learning
- policy evaluation
- function approximation
- td learning
- model free
- evaluation function
- reinforcement learning algorithms
- temporal difference learning
- action selection
- monte carlo
- function approximators
- actor critic
- approximation methods
- temporal difference methods
- step size
- machine learning
- markov decision processes
- policy iteration
- supervised learning
- state space
- dynamic programming
- dynamical systems
- multi agent
- learning algorithm
- optimal control
- policy gradient
- cost function
- active learning