GUBS: a Utility-Based Semantic for Goal-Directed Markov Decision Processes.
Valdinei FreireKarina Valdivia DelgadoPublished in: AAMAS (2017)
Keyphrases
- markov decision processes
- goal directed
- state space
- finite state
- dynamic programming
- reinforcement learning
- autonomous robots
- policy iteration
- optimal policy
- risk sensitive
- transition matrices
- reachability analysis
- finite horizon
- decision theoretic planning
- average reward
- action space
- infinite horizon
- reinforcement learning algorithms
- average cost
- planning under uncertainty
- model based reinforcement learning
- markov decision process
- factored mdps
- decision processes
- decision making
- search space
- state abstraction
- multi agent
- decision diagrams
- semi markov decision processes
- multistage
- fixed point