跳到正文
返回论文列表
cs.LG提交于 已译

SplitReAlign:分区感知的表示重对齐用于拆分联邦遗忘

SplitReAlign: Partition-Aware Representation Realignment for Split Federated Unlearning

Jia Chen · Yecheng Lv · Xinrong Hu · Jun Bai

中文摘要

拆分联邦学习(SFL)将神经网络分布在客户端与服务器上,使类特定知识在双方表示间耦合。这种耦合使拆分联邦遗忘(SFU)变得复杂:按类遗忘需在异构表示空间与非对称更新约束下移除目标知识并保留剩余效用。SplitReAlign 是一个面向按类 SFU 的分区感知表示重对齐框架。它构建端点特定的判别子空间,通过共享语义锚点协调,选择性重对齐被遗忘的表示,同时惩罚活跃子空间之外的改动。为适应非对称更新约束,SplitReAlign 在仅客户端更新时利用穿过冻结服务器的梯度,在仅服务器更新时使用固定的客户端特征统计量,而联合更新则同时优化两个分区。在异构客户端分布下的 CIFAR-10 与 CIFAR-100 实验中,SplitReAlign 在六种设置下与重训练的平均差距均为所有对比方法中最低。值得注意的是,仅服务器更新在两个数据集上都与重训练最为接近,凸显了受限更新下分区感知引导的有效性。

关键要点

  1. 01问题:SFL 中类知识跨客户端与服务器表示耦合,使按类遗忘在异构表示与非对称更新约束下难以兼顾遗忘与保留。
  2. 02方法:构建端点特定判别子空间,经共享语义锚点协调,选择性重对齐被遗忘表示并惩罚子空间外改动。
  3. 03方法:针对非对称更新,仅客户端用冻结服务器梯度,仅服务器用固定客户端特征统计量,联合更新优化两分区。
  4. 04结果:CIFAR-10 与 CIFAR-100 异构客户端分布下,六种设置中与重训练平均差距最低。
  5. 05结果:仅服务器更新在两数据集上最接近重训练,表明受限更新下分区感知引导有效。

解读

尚无解读。

原始英文摘要

arXiv:2611.25301v1 Announce Type: new Abstract: Split federated learning (SFL) distributes neural networks across clients and a server, coupling class-specific knowledge across their representations. This coupling complicates split federated unlearning (SFU), where class-wise forgetting must remove target knowledge while preserving retained utility under heterogeneous representation spaces and asymmetric update constraints. We propose SplitReAlign, a partition-aware representation realignment framework for class-wise SFU. It constructs endpoint-specific discriminative subspaces coordinated through a shared semantic anchor, selectively realigning forgotten representations while penalizing changes outside the active subspaces. To accommodate asymmetric update constraints, SplitReAlign leverages gradients through frozen servers for Client-only updates and fixed client-feature statistics for Server-only guidance, while Joint updates optimize both partitions. Experiments on CIFAR-10 and CIFAR-100 under heterogeneous client distributions demonstrate that SplitReAlign achieves the lowest average gap to retraining among all compared methods across six settings. Notably, Server-only achieves the closest agreement with retraining on both datasets, highlighting the effectiveness of partition-aware guidance under restricted updates.

同方向论文 · cs.LG

查看全部 →