解决异构设备联邦学习中个性化与通用性矛盾，提升跨设备推理精度。
HARMONY: Bridging the Personalization-Generalization Gap by Mitigating Representation Skew in Heterogeneous Split Federated Learning
- 通过模拟多样化客户端结构，动态优化个性化特征提取器。
- 服务器端对比学习对齐特征，使跨设备特征分布一致，提升外分布预测准确率。
- 支持不同设备架构，兼顾低延迟与高精度，适合资源受限的移动端应用。
移动设备面临多样化的资源约束和非独立同分布（non-IID）数据分布，需在本地快速完成对内分布（ID）类别的推理，并在需要时通过远程支持处理客户特定的外分布（OOD）类别。混合分层联邦学习（Hybrid SFL）结合了个性化客户端前端（支持早期退出）与通用服务器后端，实现精度与成本的平衡。然而，在客户端架构异构条件下，现有方法存在表示偏移问题：定制化特征提取器生成的特征在共享空间中无法对齐，导致负责OOD预测的服务器模型性能急剧下降。本文提出HARMONY，首个支持异构客户端架构的混合分层联邦学习框架。HARMONY改进元学习机制，模拟跨参数与架构的多样化提取器并实现个性化；通过服务器端对比学习对齐提取特征，既不牺牲客户端个性化能力，也无需共享原始标签。在多个数据集与模型族上，相较当前最优方法，HARMONY在无OOD场景下测试准确率提升最高达43.0%，有OOD场景下提升28.3%，同时保持可接受的延迟。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。