MARLIN: Soft Actor-Critic based Reinforcement Learning for Congestion Control in Real Networks.
Raffaele GallieraAlessandro MorelliRoberto FrontedduNiranjan SuriPublished in: CoRR (2023)
Keyphrases
- actor critic
- congestion control
- reinforcement learning
- end to end
- temporal difference
- policy gradient
- reinforcement learning algorithms
- approximate dynamic programming
- optimal control
- function approximation
- neuro fuzzy
- packet loss
- window size
- gradient method
- policy iteration
- quality of service
- state space
- rate control
- markov decision processes
- average reward
- network bandwidth
- multi agent
- model free
- action selection
- partially observable
- learning algorithm
- partially observable markov decision processes
- machine learning
- evaluation function
- supervised learning
- dynamical systems
- cost function
- convergence rate