Improved Memory-Bounded Dynamic Programming for Decentralized POMDPs
Sven SeukenShlomo ZilbersteinPublished in: CoRR (2012)
Keyphrases
- dynamic programming
- dec pomdps
- reinforcement learning
- partially observable markov decision processes
- infinite horizon
- multi agent
- state space
- decision theoretic
- theoretical justification
- linear programming
- markov decision processes
- optimal policy
- memory requirements
- partially observable
- markov decision problems
- continuous state
- single agent
- policy search
- distributed constraint optimization
- coarse to fine
- computing power
- memory usage
- distributed systems
- optimal control
- data sets
- single machine
- greedy algorithm
- main memory
- optimal solution
- machine learning