CVaR Optimization for MDPs: Existence and Computation of Optimal Policies.
Rui DingEugene A. FeinbergPublished in: SIGMETRICS Perform. Evaluation Rev. (2022)
Keyphrases
- optimal policy
- markov decision processes
- finite horizon
- reinforcement learning
- state space
- decision problems
- average reward
- dynamic programming
- policy iteration
- finite state
- average cost
- infinite horizon
- markov decision process
- robust optimization
- state dependent
- multistage
- semi markov decision processes
- sufficient conditions
- markov decision problems
- stationary policies
- long run
- reward function
- discount factor
- initial state
- dynamic programming algorithms
- reinforcement learning algorithms
- partially observable markov decision processes
- discounted reward
- average reward reinforcement learning
- bayesian reinforcement learning
- sample path
- data mining
- policy evaluation
- inventory level
- partially observable
- temporal difference
- production system
- machine learning