arXiv 2605.072112026-05-08cs.LGcs.AI

HARMONY: 通过减轻异构分裂联邦学习中的表示偏移来弥合个性化与泛化之间的鸿沟

HARMONY: Bridging the Personalization-Generalization Gap by Mitigating Representation Skew in Heterogeneous Split Federated Learning

Department of ECE and INMC, Seoul National University, Seoul, South Korea（首尔国立大学电子与信息科学系及网络信息中心，首尔，韩国）；Graduate School of Data Science, Seoul National University, Seoul, South Korea（首尔国立大学数据科学研究生院，首尔，韩国）；Department of Computer Science, University of Colorado Boulder, CO, USA（科罗拉多大学博尔德分校计算机科学系，科罗拉多州，美国）

Jiseok Youn, You Rim Choi, Goodsol Lee, Sangtae Ha, Hyung-Sin Kim, Saewoong Bahk

备注：: 7 pages (except references), 5 figures

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

HARMONY通过减轻异构分裂联邦学习中的表示偏移，实现个性化与泛化之间的平衡，提升OOD预测精度，同时保持低延迟。

摘要：

移动设备面临多样化的资源限制和非独立同分布的数据类别分布，需要快速在设备上进行本地分布（ID）类推断，并在需要时通过远程支持处理客户端特定的分布外（OOD）类推断。混合分裂联邦学习（Hybrid SFL）将个性化客户端侧前端（支持早期退出）与通用服务器侧后端结合，平衡准确性和成本。然而，在客户端架构异质性的情况下，现有混合SFL面临表示偏移问题，即定制提取器的特征无法在共享空间对齐，导致负责OOD预测的服务器模型性能急剧下降。我们提出了HARMONY，第一个支持异构客户端架构的混合SFL框架。HARMONY修改元学习以模拟不同参数和架构下的提取器，并学习个性化。为减轻表示偏移，HARMONY在服务器端进行对比学习以对齐提取特征，既不牺牲客户端的个性化也不共享原始标签。与多个数据集和模型家族的最新状态相比，HARMONY在无/有OOD情况下分别将测试精度提高高达43.0%/28.3%，同时保持可接受的延迟。

英文摘要：

Mobile devices face diverse resource constraints and non-IID data class distributions, requiring fast on-device inference for local in-distribution (ID) classes and on-demand remote support for client-specific out-of-distribution (OOD) classes. Hybrid split federated learning (Hybrid SFL) couples personalized client-side front ends (supporting early exit) with a generalized server-side backend for fallback inference, balancing accuracy and cost. However, under client architectural heterogeneity, the existing hybrid SFL suffers from representation skew, where features from customized extractors fail to align in the shared space, leading to a sharp degradation in the server model responsible for OOD prediction. We propose HARMONY, the first hybrid SFL framework to support heterogeneous client architectures. HARMONY modifies meta-learning to simulate diverse extractors across parameters and architectures, and to learn to personalize. To mitigate representation skew, HARMONY conducts server-side contrastive learning to align extracted features, neither sacrificing clients' personalization nor sharing raw labels. Compared to the state of the art across multiple datasets and model families, HARMONY improves test accuracy by up to 43.0%/28.3% without/with OOD, respectively, while maintaining acceptable latency.

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。