Online Reinforcement Learning of Optimal Threshold Policies for Markov Decision Processes.
Arghyadip RoyVivek S. BorkarAbhay KarandikarPrasanna ChaporkarPublished in: CoRR (2019)
Keyphrases
- markov decision processes
- optimal policy
- reinforcement learning
- average cost
- stationary policies
- dynamic programming
- total reward
- policy iteration algorithm
- average reward
- discounted reward
- markov decision process
- finite horizon
- state space
- policy iteration
- action sets
- reward function
- reinforcement learning algorithms
- control policies
- discount factor
- finite state
- expected reward
- control policy
- decision processes
- state and action spaces
- transition matrices
- partially observable
- markov decision problems
- decision problems
- planning under uncertainty
- partially observable markov decision processes
- infinite horizon
- optimality criterion
- approximate dynamic programming
- decentralized control
- action space
- model based reinforcement learning
- decision theoretic planning
- machine learning
- factored mdps
- optimal control
- initial state
- macro actions
- multi agent
- model free
- long run
- function approximation
- rl algorithms
- linear programming
- temporal difference