Normative Reasoning with an Adaptive Self-interested Agent Model Based on Markov Decision Processes.
Moser Silva FagundesHolger BillhardtSascha OssowskiPublished in: IBERAMIA (2010)
Keyphrases
- markov decision processes
- multi agent systems
- partially observable
- reward function
- reasoning process
- markov decision process
- multi agent
- state abstraction
- optimal policy
- reinforcement learning
- state space
- interval estimation
- action space
- discounted reward
- finite state
- expected reward
- policy iteration
- planning under uncertainty
- decision theoretic planning
- multiagent systems
- reinforcement learning algorithms
- model based reinforcement learning
- finite horizon
- reachability analysis
- knowledge base
- dynamic programming
- multiple agents
- infinite horizon
- average cost
- decision processes
- factored mdps
- action sets
- transition matrices
- decision theoretic
- bdi agents
- average reward
- model free
- decision making
- dynamic environments
- state and action spaces
- total reward
- single agent
- function approximation
- action selection
- semi markov decision processes
- learning agent
- plan execution