MR-ARL: Model Reference Adaptive Reinforcement Learning for Robustly Stable On-Policy Data-Driven LQR.
Marco BorghesiAlessandro BossoGiuseppe NotarstefanoPublished in: CoRR (2024)
Keyphrases
- data driven
- model reference adaptive
- reinforcement learning
- optimal policy
- policy search
- optimal control
- action selection
- partially observable environments
- markov decision process
- function approximators
- actor critic
- reinforcement learning problems
- function approximation
- markov decision problems
- state space
- policy iteration
- reinforcement learning algorithms
- control policy
- action space
- policy gradient
- infinite horizon
- state and action spaces
- partially observable
- control policies
- rl algorithms
- reward function
- image registration
- markov decision processes
- magnetic resonance
- mr images
- partially observable markov decision processes
- inverse reinforcement learning
- partially observable domains
- policy evaluation
- decision problems
- continuous state spaces
- temporal difference
- transfer learning
- dynamic programming
- machine learning
- continuous state
- average reward
- control problems
- medical images
- mr imaging
- state dependent
- model free reinforcement learning
- average cost
- model free
- long run
- image data
- policy gradient methods
- agent receives
- multi agent
- linear quadratic
- finite state
- partially observable markov decision process
- reinforcement learning methods