Communication-based cooperative tasks: how the language expressiveness affects reinforcement learning.
Jacopo TalaminiEric MedvetAlberto BartoliPublished in: SAC (2019)
Keyphrases
- cooperative
- reinforcement learning
- transfer learning
- multi agent
- language learning
- multi agent systems
- fully distributed
- expressive power
- programming language
- distributed control
- communication systems
- object oriented
- state space
- natural language
- communication networks
- function approximation
- communication protocol
- information sharing
- multi agent reinforcement learning
- solve complex tasks
- human communication
- agent communication
- markov decision processes
- previously learned
- model free
- communication channels
- neural network
- natural language processing
- knowledge representation
- learning algorithm