Connectivity-Influence Weighting: A Structural Bias for Graph-Based MARL
Graph neural networks used for multi-agent reinforcement learning typically treat every edge of the communication graph as equally important. This paper introduces …
Salah Chegri