Agentic联邦学习：基于规则的客户端与服务器智能体实现自适应训练
原标题：Agentic Federated Learning: Rule-Based Client and Server Agents for Adaptive Training
AI 摘要
该论文提出了一种 Agentic Federated Learning（AFL）框架，在客户端和服务器端分别引入轻量级基于规则的自主智能体：客户端智能体（CSA）动态调整本地训练参数、控制参与并评估更新可靠性，服务器端编排智能体（SSOA）进行质量感知的客户端选择与自适应聚合。在 CIFAR-10 数据集上针对 IID、非 IID 和噪声客户端设置的实验表明，AFL 在分类准确率、收敛速度、抗噪声鲁棒性和通信效率上均优于 FedAvg 和 FedProx 等基线，消融实验验证了两个智能体的互补贡献。
正文节选
Agentic Federated Learning: Rule-Based Client and Server Agents for Adaptive Training Abstract Federated Learning (FL) enables collaborative model training across distributed clients without sharing raw data, making it suitable for privacy-sensitive applications such as healthcare, finance, and edge intelligence. However, conventional FL approaches rely on static client participation and fixed aggregation strategies, which limits their effectiveness under non-IID data distributions, heterogeneou