Transfer reinforcement learning for fault-tolerant control by re-using optimal policies.
Ibrahim AhmedMarcos Quiñones-GrueiroGautam BiswasPublished in: SysTol (2021)
Keyphrases
- optimal policy
- reinforcement learning
- fault tolerant control
- transfer learning
- markov decision processes
- nonlinear systems
- stability analysis
- decision problems
- control system
- state space
- infinite horizon
- finite horizon
- dynamic programming
- finite state
- markov decision process
- long run
- adaptive control
- control scheme
- control policies
- average reward
- state dependent
- reward function
- average cost
- sufficient conditions
- multi agent
- initial state
- partially observable markov decision processes
- average reward reinforcement learning
- policy iteration
- control law
- reinforcement learning algorithms
- function approximation
- dynamic environments
- expert systems
- learning algorithm
- total reward
- markov decision problems
- action space
- temporal difference