Symbolic Dynamic Programming for Risk-sensitive Markov Decision Process with limited budget.
Daniel A. M. MoreiraKarina Valdivia DelgadoLeliane Nunes de BarrosPublished in: AAMAS (2018)
Keyphrases
- risk sensitive
- markov decision process
- limited budget
- markov decision processes
- dynamic programming
- optimal policy
- optimal control
- infinite horizon
- state space
- finite horizon
- markov decision problems
- control policies
- policy iteration
- weighted distance
- average cost
- reinforcement learning
- finite state
- multistage
- action space
- reinforcement learning algorithms
- partially observable
- linear programming
- reward function
- long run
- average reward
- utility function