A Surprisingly Simple Continuous-Action POMDP Solver: Lazy Cross-Entropy Search Over Policy Trees.
Marcus HörgerHanna KurniawatiDirk P. KroeseNan YePublished in: AAAI (2024)
Keyphrases
- cross entropy
- partially observable markov decision processes
- policy search
- continuous action
- continuous state
- optimal policy
- search space
- reinforcement learning
- dynamic programming
- state space
- decision problems
- partially observable
- finite state
- markov decision processes
- dynamical systems
- log likelihood
- action space
- ranking functions
- planning problems
- neural network
- markov chain
- maximum likelihood
- information retrieval
- machine learning