Optimal policies for some n-dimensional singular stochastic control problems.
Lukasz KrukPublished in: CDC (2001)
Keyphrases
- control problems
- optimal policy
- stochastic control
- reinforcement learning
- optimal control
- dynamic programming
- markov decision processes
- state dependent
- infinite horizon
- state space
- queueing systems
- decision problems
- control policies
- average cost
- finite horizon
- long run
- function approximation
- average reward
- finite state
- continuous state spaces
- partially observable markov decision processes
- multistage
- dynamic programming algorithms
- initial state
- markov decision process
- adaptive control
- average reward reinforcement learning
- markov decision problems
- serial inventory systems
- brownian motion
- sufficient conditions
- asymptotically optimal
- temporal difference
- model free
- function approximators
- lost sales
- learning algorithm
- reinforcement learning algorithms
- base stock policies
- action selection
- inventory level
- steady state
- control system