Concurrent MDPs with Finite Markovian Policies.
Peter BuchholzDimitri ScheftelowitschPublished in: MMB (2020)
Keyphrases
- optimal policy
- markov decision processes
- state and action spaces
- markov decision problems
- temporally extended
- decision theoretic planning
- markov decision process
- semi markov decision process
- hierarchical reinforcement learning
- average cost
- stationary policies
- reinforcement learning
- reward function
- policy search
- finite number
- state space
- average reward
- dynamic programming
- decision processes
- factored mdps
- finite horizon
- finite state
- action space
- infinite horizon
- policy iteration
- discounted reward
- long run
- planning under uncertainty
- decision problems
- expected reward
- factored markov decision processes
- control policies
- linear programming
- probabilistic planning
- utility function
- sufficient conditions
- search algorithm