Approximate Planning in POMDPs with Macro-Actions.
Georgios TheocharousLeslie Pack KaelblingPublished in: NIPS (2003)
Keyphrases
- macro actions
- markov decision processes
- reinforcement learning
- state space
- markov decision problems
- planning domains
- partially observable
- optimal policy
- partially observable markov decision processes
- point based value iteration
- predictive model
- belief state
- function approximation
- finite state
- policy iteration
- temporally extended
- plan library
- learned knowledge
- planning under uncertainty
- temporal difference
- continuous state
- neural network
- model free
- linear programming
- dynamic programming
- machine learning