Coarticulation: an approach for generating concurrent plans in Markov decision processes.
Khashayar RohanimaneshSridhar MahadevanPublished in: ICML (2005)
Keyphrases
- markov decision processes
- planning under uncertainty
- macro actions
- decision theoretic planning
- temporally extended
- state space
- finite state
- reinforcement learning
- optimal policy
- reinforcement learning algorithms
- transition matrices
- policy iteration
- dynamic programming
- reachability analysis
- risk sensitive
- average reward
- action space
- factored mdps
- finite horizon
- model based reinforcement learning
- markov decision process
- infinite horizon
- average cost
- probabilistic planning
- decision processes
- action sets
- search space
- real time dynamic programming
- data mining
- state abstraction
- initial state
- reward function
- heuristic search
- monte carlo
- machine learning