arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2512.03336cs.LGcs.AIstat.ML
收藏

单轮可扩展的分析联邦学习

Single-Round Scalable Analytic Federated Learning

University of Texas at Austin(德克萨斯大学奥斯汀分校);Federal University of Rio de Janeiro(里约热内卢联邦大学);Instituto de Telecomunicações, Porto(波尔图电信研究所)

Alan T. L. Bacellar, Mustafa Munir, Felipe M. G. França, Priscila M. V. Lima, Radu Marculescu, Lizy K. John

AI总结:

本文提出SAFLe框架,通过引入分桶特征结构头部和稀疏分组嵌入,实现非线性可扩展性,证明其等价于高维线性回归,从而在单轮分析联邦学习中实现高效可扩展的联邦视觉解决方案。

摘要:

联邦学习(FL)面临两个关键挑战:高通信开销和异构(非iid)数据上的性能崩溃。分析联邦学习(AFL)提供了一种单轮、数据分布不变的解决方案,但受限于线性模型。后续非线性方法如DeepAFL在准确率上恢复但牺牲了单轮优势。本文打破这一权衡,提出SAFLe框架,通过引入分桶特征结构头部和稀疏分组嵌入实现可扩展的非线性表达性。证明该非线性架构在数学上等价于高维线性回归。这一关键等价性允许SAFLe通过AFL的单次、不变聚合法则进行求解。实证结果表明,SAFLe在分析联邦学习中建立了新的状态-of-the-art,显著优于线性AFL和多轮DeepAFL,在所有基准测试中均表现出更高的准确率,展示了高效且可扩展的联邦视觉解决方案。

英文摘要:

Federated Learning (FL) is plagued by two key challenges: high communication overhead and performance collapse on heterogeneous (non-IID) data. Analytic FL (AFL) provides a single-round, data distribution invariant solution, but is limited to linear models. Subsequent non-linear approaches, like DeepAFL, regain accuracy but sacrifice the single-round benefit. In this work, we break this trade-off. We propose SAFLe, a framework that achieves scalable non-linear expressivity by introducing a structured head of bucketed features and sparse, grouped embeddings. We prove this non-linear architecture is mathematically equivalent to a high-dimensional linear regression. This key equivalence allows SAFLe to be solved with AFL's single-shot, invariant aggregation law. Empirically, SAFLe establishes a new state-of-the-art for analytic FL, significantly outperforming both linear AFL and multi-round DeepAFL in accuracy across all benchmarks, demonstrating a highly efficient and scalable solution for federated vision.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑