一项新的研究论文提出，“边缘稳定性”（EoS）现象阻碍了SCAFFOLD算法在联邦优化中的性能。尽管SCAFFOLD在处理数据异质性方面具有理论优势，但实际上它通常表现不如更简单的FedAvg。研究表明，这是因为EoS动力学（以渐进锐化为特征）会损害SCAFFOLD准确估计全局梯度的能力，尤其是在高锐度水平下。 AI
影响 这项研究可能导致改进的联邦学习算法，更好地处理数据异质性。
排序理由 在arXiv上发表的研究论文，详细介绍了关于联邦优化算法的新发现。
在 Hugging Face Daily Papers 阅读 →
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
arXiv:2608.25873v1 Announce Type: new Abstract: In federated learning, it is well known that heterogeneous data can (in theory) slow down optimization, and much effort has been directed at designing optimization algorithms that are unaffected by data heterogeneity, such as the SC…
In federated learning, it is well known that heterogeneous data can (in theory) slow down optimization, and much effort has been directed at designing optimization algorithms that are unaffected by data heterogeneity, such as the SCAFFOLD algorithm. Yet, despite strong theoretica…