A Variational Perturbative Approach to Planning in Graph-based Markov Decision Processes.
Dominik LinznerHeinz KoepplPublished in: CoRR (2019)
Keyphrases
- markov decision processes
- macro actions
- planning under uncertainty
- decision theoretic planning
- partially observable
- state space
- optimal policy
- finite state
- reinforcement learning
- transition matrices
- dynamic programming
- probabilistic planning
- policy iteration
- heuristic search
- reinforcement learning algorithms
- partially observable markov decision processes
- infinite horizon
- reachability analysis
- finite horizon
- planning problems
- ai planning
- average cost
- action space
- state and action spaces
- model based reinforcement learning
- action sets
- markov decision process
- blocks world
- factored mdps
- machine learning
- optical flow
- state abstraction
- decision processes
- reward function
- decision theoretic
- dynamical systems
- least squares
- average reward
- action selection