Flexible approximation of structured interactions in decentralized Markov decision processes.
Stefan J. WitwickiEdmund H. DurfeePublished in: AAMAS (2) (2009)
Keyphrases
- markov decision processes
- decentralized control
- state space
- optimal policy
- reinforcement learning
- finite state
- policy iteration
- dynamic programming
- reachability analysis
- transition matrices
- infinite horizon
- factored mdps
- policy evaluation
- policy iteration algorithm
- action sets
- average reward
- markov decision process
- partially observable
- action space
- decision theoretic planning
- risk sensitive
- reinforcement learning algorithms
- planning under uncertainty
- average cost
- decision processes
- approximation methods
- cooperative
- stationary policies
- state abstraction
- reward function
- state and action spaces
- finite number
- function approximation
- semi markov decision processes
- real time dynamic programming
- machine learning