Design and evaluation of norm-aware agents based on Normative Markov Decision Processes.
Moser Silva FagundesSascha OssowskiJesús CerquidesPablo NoriegaPublished in: Int. J. Approx. Reason. (2016)
Keyphrases
- markov decision processes
- multi agent systems
- state space
- multi agent
- finite state
- normative systems
- decentralized control
- optimal policy
- policy iteration
- reinforcement learning
- reachability analysis
- planning under uncertainty
- reinforcement learning algorithms
- dynamic programming
- multiple agents
- transition matrices
- factored mdps
- stochastic games
- decision processes
- action space
- finite horizon
- long run
- single agent
- decision theoretic planning
- expected reward
- cooperative
- infinite horizon
- decision theoretic
- average reward
- partially observable
- least squares
- reward function
- learning algorithm