Markov Decision Processes with Threshold Based Piecewise Linear Optimal Policies.
Tomaso ErsegheAndrea ZanellaClaudio G. CodemoPublished in: IEEE Wirel. Commun. Lett. (2013)
Keyphrases
- piecewise linear
- markov decision processes
- optimal policy
- dynamic programming
- state space
- finite state
- decision problems
- finite horizon
- reinforcement learning
- average reward
- infinite horizon
- policy iteration
- multistage
- state dependent
- long run
- average cost
- decision processes
- sufficient conditions
- markov decision process
- partially observable
- initial state
- reinforcement learning algorithms
- linear programming
- reward function
- action space
- decision diagrams
- sample path
- partially observable markov decision processes
- data mining
- hyperplane
- multi agent
- lost sales
- state and action spaces
- total reward
- discounted reward
- semi markov decision processes