RLlib Flow: Distributed Reinforcement Learning is a Dataflow Problem.
Eric LiangZhanghao WuMichael LuoSven MikaJoseph E. GonzalezIon StoicaPublished in: NeurIPS (2021)
Keyphrases
- reinforcement learning
- multi agent
- distributed systems
- distributed environment
- lightweight
- computer networks
- cooperative
- peer to peer
- optimal control
- function approximation
- learning algorithm
- fault tolerant
- reinforcement learning methods
- data flow
- communication cost
- flow field
- markov decision processes
- mobile agents
- response time