Scalable grid-based approximation algorithms for partially observable Markov decision processes.
Can KavakliogluMucahit CevikPublished in: Concurr. Comput. Pract. Exp. (2022)
Keyphrases
- approximation algorithms
- partially observable markov decision processes
- np hard
- finite state
- decision problems
- reinforcement learning
- dynamical systems
- belief space
- special case
- dynamic programming
- belief state
- state space
- optimal policy
- worst case
- partially observable stochastic games
- vertex cover
- multi agent
- partially observable markov
- planning problems
- markov decision processes
- approximate solutions
- partially observable
- approximation ratio
- lower bound
- undirected graph
- infinite horizon
- average reward
- dynamic environments
- computational complexity
- machine learning