Anderson Acceleration for Partially Observable Markov Decision Processes: A Maximum Entropy Approach.
Mingyu ParkJaeuk ShinInsoon YangPublished in: CoRR (2022)
Keyphrases
- maximum entropy
- partially observable markov decision processes
- finite state
- reinforcement learning
- maximum entropy principle
- dynamical systems
- decision problems
- optimal policy
- belief state
- state space
- dynamic programming
- markov models
- partially observable stochastic games
- markov decision processes
- partially observable
- conditional random fields
- transformation based learning
- multi agent
- planning problems
- infinite horizon
- decision trees
- markov chain
- minimum cross entropy
- approximate solutions
- point based value iteration
- computational complexity
- objective function
- special case
- energy function
- maximum likelihood