Deep learning

Deep active inference agents using Monte-Carlo methods

Active Inference just got real, thanks to Monte Carlo trees and information theory.

Relational forward models for multi-agent learning

Using graph neural networks to understand and improve multi-agent RL.