TL;DR提出了 Individually Inferred Communication 模型,通过因果推断学习通信先验,并利用前馈神经网络实现代理之间的通信。该模型成功地减少了通信开销并提高了多种多代理合作场景中的表现。
Abstract
Communication lays the foundation for human cooperation. It is also crucial
for multi-agent cooperation. However, existing work focuses on broadcast
communication, which is not only impractical but also leads to
本文提出了 Individualized Controlled Continuous Communication Model (IC3Net),在多智能体协作、半协作与竞争环境下,通过门控机制控制持续传输,并使用个性化奖励来提高性能和可扩展性,修正学分分配问题。实验结果证实,IC3Net 网络比基准网络在不同场景下具有更好的训练效率和收敛率,智能体基于场景和可盈利性学会如何传输信息。