Approximate dynamic programming for stochastic linear control problems on compact state spaces.
Stefan WörnerMarco LaumannsRico ZenklusenApostolos FertisPublished in: Eur. J. Oper. Res. (2015)
Keyphrases
- control problems
- approximate dynamic programming
- reinforcement learning
- stochastic dynamic programming
- state space
- continuous state spaces
- optimal control
- reinforcement learning algorithms
- dynamic programming
- stochastic control
- linear program
- markov decision processes
- average cost
- function approximation
- adaptive control
- control policies
- model free
- optimal policy
- function approximators
- control policy
- action space
- policy iteration
- hamilton jacobi bellman
- temporal difference
- learning algorithm
- markov decision process
- partially observable
- continuous state
- step size
- dynamical systems
- sufficient conditions
- markov chain
- computational complexity
- data mining
- real time