Gradual-Impulsive Control for Continuous-Time Markov Decision Processes with Total Undiscounted Costs and Constraints: Linear Programming Approach via a Reduction Method.
Alexey PiunovskiyYi ZhangPublished in: SIAM J. Control. Optim. (2022)
Keyphrases
- markov decision processes
- reduction method
- linear programming
- dynamic programming
- optimal control
- average cost
- state space
- policy iteration
- optimal policy
- infinite horizon
- finite state
- reinforcement learning
- average reward
- reinforcement learning algorithms
- transition matrices
- selection algorithm
- finite horizon
- dynamical systems
- markov decision problems
- partially observable
- linear program
- decision processes
- control strategy
- control system
- multistage
- markov chain
- stationary policies
- decision theoretic planning
- evolutionary algorithm
- factored mdps
- total reward
- real time dynamic programming
- stochastic games
- action space
- expected cost
- feature selection
- computational complexity
- markov decision process
- partially observable markov decision processes
- data structure
- long run
- stochastic shortest path