Modeling of route planning system based on Q value-based dynamic programming with multi-agent reinforcement learning algorithms.
Mortaza Zolfpour ArokhloAli SelamatSiti Zaiton Mohd HashimHossein AfkhamiPublished in: Eng. Appl. Artif. Intell. (2014)
Keyphrases
- function approximators
- reinforcement learning algorithms
- reinforcement learning
- dynamic programming
- multi agent
- function approximation
- state space
- reinforcement learning problems
- markov decision processes
- temporal difference
- model free
- policy search
- reinforcement learning methods
- multi agent environments
- markov games
- eligibility traces
- learning algorithm
- machine learning
- optimal policy
- linear programming
- cooperative
- multiagent reinforcement learning
- single agent
- stochastic games
- multi agent systems
- convergence speed
- dynamic environments
- random walk
- markov chain
- training data
- reward shaping