FedRAPT: federated representation-aligned prototypical contrastive learning
4 affiliations
- 1Department of Computer Engineering, Sunchon National University
- 2Interdisciplinary Program in IT-Bio Convergence System, Sunchon National University
- 3Department of Information Communication and Multimedia Engineering, Sunchon National University
- 4Department of Computer Education, Gyeongkuk National University
Abstract
Federated learning (FL) is a distributed learning paradigm that enables collaborative model training among multiple clients without sharing raw data. FL has recently attracted attention in privacy-sensitive environments, such as mobile sensing and human activity recognition (HAR). However, in real-world environments, data heterogeneity among clients causes non-independent and identically distributed (non-IID) conditions, which induce class-level representation inconsistency and representation drift, thereby degrading the global model performance and training stability. Existing studies have mainly focused on either global alignment or local discrimination, making it difficult to simultaneously reflect class structural consistency and sample discriminability in non-IID environments. To solve this problem, we propose Federated Representation-Aligned Prototypical Contrastive Learning (FedRAPT). FedRAPT introduced an information-noise-contrastive estimation (InfoNCE)-based Cross-client Contrastive Representation Alignment (CCRA) module to jointly optimize class-level alignment and sample-level discrimination using a unified objective function. In addition, an encoder classifier decoupled structure and an exponential moving average (EMA) based prototype update strategy were designed to induce stable representation learning, even in highly heterogeneous data distributions. Experimental results on the WISDM, UCI HAR, and MotionSense datasets showed that the proposed method achieved accuracies of 96.30%, 96.13%, and 94.00% accuracy, respectively, and maintained an accuracy of approximately 94% even in a strong non-IID environment. In addition, improvements in the loss convergence and forgetting rate confirm that the proposed method effectively alleviates performance degradation and training instability caused by data heterogeneity. These results demonstrate that FedRAPT achieves robust and consistent representation learning under non-IID conditions.
Related
- STCP-Net: A few-shot learning approach based on spectral-temporal convolution and an improved prototypical network for ECG classification
- Contrastive adversarial mix-up learning for unsupervised cross-modality image segmentation
- Contrastive learning and advanced bundle metrics outperform fractional anisotropy for Alzheimer’s disease diffusion MRI prediction: findings from the Beyond FA challenge
- Deep Learning for Multi‐Disease OCT Classification in Clinical Retina Practice Using Attention Mechanisms and Contrastive Learning
- PRInter: an inductive multimodal heterogeneous network embedding framework with contrastive learning for lncRNA-protein interaction prediction
BibTeX
@article{seo2026fedrapt,
title = {{FedRAPT: federated representation-aligned prototypical contrastive learning}},
author = {Su-Bin Seo and Jong-Hoon Kim and Chang-Sun Shin and Han-Sung Lee and Chun-Bo Sim and Se-Hoon Jung},
journal = {Scientific Reports},
year = {2026},
month = jul,
doi = {10.1038/s41598-026-62486-3},
url = {https://doi.org/10.1038/s41598-026-62486-3},
note = {CC BY},
}
RIS
TY - JOUR
TI - FedRAPT: federated representation-aligned prototypical contrastive learning
AU - Su-Bin Seo
AU - Jong-Hoon Kim
AU - Chang-Sun Shin
AU - Han-Sung Lee
AU - Chun-Bo Sim
AU - Se-Hoon Jung
PY - 2026
DA - 2026/07/01
JO - Scientific Reports
DO - 10.1038/s41598-026-62486-3
UR - https://doi.org/10.1038/s41598-026-62486-3
AB - Federated learning (FL) is a distributed learning paradigm that enables collaborative model training among multiple clients without sharing raw data. FL has recently attracted attention in privacy-sensitive environments, such as mobile sensing and human activity recognition (HAR). However, in real-world environments, data heterogeneity among clients causes non-independent and identically distributed (non-IID) conditions, which induce class-level representation inconsistency and representation drift, thereby degrading the global model performance and training stability. Existing studies have mainly focused on either global alignment or local discrimination, making it difficult to simultaneously reflect class structural consistency and sample discriminability in non-IID environments. To solve this problem, we propose Federated Representation-Aligned Prototypical Contrastive Learning (FedRAPT). FedRAPT introduced an information-noise-contrastive estimation (InfoNCE)-based Cross-client Contrastive Representation Alignment (CCRA) module to jointly optimize class-level alignment and sample-level discrimination using a unified objective function. In addition, an encoder classifier decoupled structure and an exponential moving average (EMA) based prototype update strategy were designed to induce stable representation learning, even in highly heterogeneous data distributions. Experimental results on the WISDM, UCI HAR, and MotionSense datasets showed that the proposed method achieved accuracies of 96.30%, 96.13%, and 94.00% accuracy, respectively, and maintained an accuracy of approximately 94% even in a strong non-IID environment. In addition, improvements in the loss convergence and forgetting rate confirm that the proposed method effectively alleviates performance degradation and training instability caused by data heterogeneity. These results demonstrate that FedRAPT achieves robust and consistent representation learning under non-IID conditions.
SN - 2045-2322
ER -