Dynamic Risk Measures for Finite-State Partially Observable Markov Decision Problems.
Jingnan FanAndrzej RuszczynskiPublished in: SIAM Conf. on Control and its Applications (2015)
Keyphrases
- finite state
- partially observable
- markov decision processes
- markov decision problems
- optimal policy
- average cost
- policy iteration
- reinforcement learning
- state space
- partially observable markov decision processes
- markov chain
- infinite horizon
- dynamic programming
- decision problems
- reward function
- reinforcement learning algorithms
- action space
- fully observable
- transition probabilities
- dynamical systems
- decision processes
- linear programming
- long run
- markov decision process
- decision theoretic
- belief state
- influence diagrams