Modified Policy Iteration for Exponential Cost Risk Sensitive MDPs.
Yashaswini MurthyMehrdad MoharramiR. SrikantPublished in: CoRR (2023)
Keyphrases
- risk sensitive
- markov decision processes
- policy iteration
- average cost
- markov decision problems
- model free
- optimal policy
- finite state
- optimal control
- factored mdps
- infinite horizon
- dynamic programming
- average reward
- reinforcement learning
- policy evaluation
- state space
- finite horizon
- markov decision process
- planning under uncertainty
- reinforcement learning algorithms
- action space
- long run
- linear programming
- fixed point
- partially observable
- initial state
- decision processes
- least squares
- reward function
- sufficient conditions
- multistage
- finite number
- total cost