More Centralized Training, Still Decentralized Execution: Multi-Agent Conditional Policy Factorization.
Jiangxing WangDeheng YeZongqing LuPublished in: ICLR (2023)
Keyphrases
- multi agent
- peer to peer
- cooperative
- multi agent systems
- peer to peer systems
- training process
- intelligent agents
- reinforcement learning
- autonomous agents
- fully distributed
- single agent
- multiagent systems
- optimal policy
- test set
- distributed environment
- low rank
- supervised learning
- pairwise
- dynamic environments
- singular value decomposition
- conditional probabilities
- multiple agents
- cognitive agents
- machine learning
- neural network