Control as Probabilistic Inference as an Emergent Communication Mechanism in Multi-Agent Reinforcement Learning.
Tomoaki NakamuraAkira TaniguchiTadahiro TaniguchiPublished in: CoRR (2023)
Keyphrases
- probabilistic inference
- multi agent reinforcement learning
- distributed control
- graphical models
- bayesian networks
- belief networks
- autonomous agents
- control architecture
- conditional probabilities
- weighted model counting
- message passing
- cooperative
- multi agent
- probabilistic model
- multi agent systems
- multi agent learning
- belief propagation
- control strategy
- knowledge acquisition
- stochastic games
- training data
- neural network