reinforcement learning

Trading off Utility, Informativeness, and Complexity in Emergent Communication

Towards Human-Agent Communication via the Information Bottleneck Principle