An exact algorithm for solving MDPs under risk-sensitive planning objectives with one-switch utility functions.
Yaxin LiuSven KoenigPublished in: AAMAS (1) (2008)
Keyphrases
- risk sensitive
- markov decision problems
- utility function
- markov decision processes
- expected utility
- linear programming
- partially observable
- decision theoretic
- state space
- reinforcement learning
- decision problems
- decision processes
- optimal policy
- decision theory
- risk neutral
- decision makers
- average cost
- planning under uncertainty
- dynamic programming
- policy iteration
- transition probabilities
- blocks world
- markov chain
- infinite horizon
- queueing networks
- finite state
- risk averse
- initial state
- risk aversion
- action space
- probability distribution
- long run
- markov decision process
- learning algorithm
- multiple objectives
- optimal control
- dynamical systems
- decision making