Sufficiency of Markov policies for continuous-time Markov decision processes and solutions to Kolmogorov's forward equation for jump Markov processes.
Eugene A. FeinbergManasa MandavaAlbert N. ShiryaevPublished in: CDC (2013)
Keyphrases
- markov processes
- markov decision processes
- markov chain
- optimal policy
- state space
- finite state
- markov process
- markov decision process
- average cost
- reward function
- partially observable markov decision processes
- transition probabilities
- stochastic processes
- stationary policies
- continuous time bayesian networks
- average reward
- reinforcement learning
- transition matrices
- markov model
- steady state
- infinite horizon
- stationary distribution
- dynamic programming
- expected reward
- decision theoretic planning
- decision problems
- stochastic process
- markov decision problems
- action sets
- total reward
- discounted reward
- policy iteration
- long run
- partially observable
- optimal solution
- random fields
- action space
- maximum likelihood
- non stationary
- state variables
- machine learning