Scholarlink

<p>The widespread adoption of mobile and wearable devices has made human activity recognition (HAR) a key component of pervasive computing, enabling applications in smart healthcare, ambient assisted living, and smart cities. However, centralized HAR models face privacy concerns and performance degradation due to data heterogeneity arising from varying user behaviors, sensor placements, device types, and environmental conditions. While federated learning (FL) offers a privacy-preserving alternative, it suffers from client drift under non-IID data distributions, limiting both personalization and generalization. In this work, we propose FedAli-PHAR, a novel personalized federated learning framework that extends the Alignment with Prototypes (ALP) layer using optimal transport (Sinkhorn-Knopp algorithm) to align embeddings with learnable local and global prototypes. The proposed approach mitigates client drift through distribution-level alignment, incorporates adaptive feature fusion via a Gated Linear Unit (GLU), and employs exponential moving average (EMA) updates for stable prototype evolution in time-series sensor data. During inference, local prototypes enable efficient on-device personalization without additional communication. Extensive experiments on heterogeneous HAR benchmarks (HHAR, RealWorld, and UCI-HAR under realistic non-IID partitioning) demonstrate that FedAli-PHAR consistently outperforms state-of-the-art personalized FL methods (FedAvg, FedProx, FedProto, MOON, and FedAli), achieving statistically significant improvements of 2%–5% in personalized accuracy/F1 scores and 3%–7% in global generalization. The framework also exhibits faster convergence and reduced performance variance, while maintaining low communication overhead. FedAli-PHAR provides a scalable and practical foundation for robust, privacy-aware HAR in pervasive computing environments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

FedAli-PHAR: personalized federated learning with optimal transport-based prototype alignment for robust human activity recognition in pervasive computing

  • 期刊论文

  • 发表日期：

    2026年7月20日

查看全文
  • R. Satheeskumar,
  • S. Oyyathevan,
  • J. Thimmiaraja,
  • M. Suresh,
  • Talatoti Ratna Kumar

摘要

The widespread adoption of mobile and wearable devices has made human activity recognition (HAR) a key component of pervasive computing, enabling applications in smart healthcare, ambient assisted living, and smart cities. However, centralized HAR models face privacy concerns and performance degradation due to data heterogeneity arising from varying user behaviors, sensor placements, device types, and environmental conditions. While federated learning (FL) offers a privacy-preserving alternative, it suffers from client drift under non-IID data distributions, limiting both personalization and generalization. In this work, we propose FedAli-PHAR, a novel personalized federated learning framework that extends the Alignment with Prototypes (ALP) layer using optimal transport (Sinkhorn-Knopp algorithm) to align embeddings with learnable local and global prototypes. The proposed approach mitigates client drift through distribution-level alignment, incorporates adaptive feature fusion via a Gated Linear Unit (GLU), and employs exponential moving average (EMA) updates for stable prototype evolution in time-series sensor data. During inference, local prototypes enable efficient on-device personalization without additional communication. Extensive experiments on heterogeneous HAR benchmarks (HHAR, RealWorld, and UCI-HAR under realistic non-IID partitioning) demonstrate that FedAli-PHAR consistently outperforms state-of-the-art personalized FL methods (FedAvg, FedProx, FedProto, MOON, and FedAli), achieving statistically significant improvements of 2%–5% in personalized accuracy/F1 scores and 3%–7% in global generalization. The framework also exhibits faster convergence and reduced performance variance, while maintaining low communication overhead. FedAli-PHAR provides a scalable and practical foundation for robust, privacy-aware HAR in pervasive computing environments.