DpFedFKP: Dynamic personalized federated learning for finger knuckle print recognition

Shuyi Li; Jianian Hu; Bob Zhang; Shanping Yu; Lifang Wu

doi:10.1016/j.patcog.2025.112966

DpFedFKP: Dynamic personalized federated learning for finger knuckle print recognition

Shuyi Li
, Jianian Hu
, Bob Zhang
, Shanping Yu^*
, Lifang Wu

^*此作品的通讯作者

网络空间安全学院

Beijing University of Technology
University of Macau

科研成果: 期刊稿件 › 文章 › 同行评审

摘要

Finger knuckle print (FKP) recognition has recently attracted significant attention owing to its low cost, high security, and discriminative characteristics. However, existing data-driven approaches, particularly deep learning methods, typically require centralized training with large-scale datasets, raising serious privacy risks. Furthermore, the substantial data heterogeneity among independent datasets often leads to training difficulties and performance degradation. To address these challenges, we propose a dynamic personalized federated learning framework, called DpFedFKP, for FKP recognition. Specifically, each client first performs local training through classification loss constraints to obtain the local model, and the central server then aggregates these local models via Federated Averaging (FedAvg) to generate a generalized global model. Subsequently, each client performs an adaptive gradient analysis that computes gradients and parameter variations between global and local models using domain-specific sub-datasets. Furthermore, we propose a novel server-side dynamic update mechanism that dynamically adjusts the local model ratios within client-specific personalized models, enabling optimal interpolation between the global and local models to achieve robust generalization and personalization. Comprehensive experiments on two widely used FKP datasets demonstrate that the proposed method has achieved significant performance improvement over the state-of-the-art techniques. Specifically, our method achieves 0.03% relative accuracy improvement on Data-fkp and 0.52% relative accuracy improvement on PolyU-fkp, with 12.35% relative reduction in equal error rate (EER) on PolyU-fkp. Furthermore, the cross-client experiments achieve up to 2.80% relative accuracy improvement and 33.48% relative EER reduction. Significantly, the proposed DpFedFKP has strong compatibility with differential privacy techniques, thereby enhancing privacy-preserving capability.

源语言	英语
文章编号	112966
期刊	Pattern Recognition
卷	174
DOI	https://doi.org/10.1016/j.patcog.2025.112966
出版状态	已出版 - 6月 2026

访问文件

10.1016/j.patcog.2025.112966

其它文件与链接

链接到 Scopus 的出版物

引用此

@article{3dcc8678517344a0ad6887665a678105,

title = "DpFedFKP: Dynamic personalized federated learning for finger knuckle print recognition",

abstract = "Finger knuckle print (FKP) recognition has recently attracted significant attention owing to its low cost, high security, and discriminative characteristics. However, existing data-driven approaches, particularly deep learning methods, typically require centralized training with large-scale datasets, raising serious privacy risks. Furthermore, the substantial data heterogeneity among independent datasets often leads to training difficulties and performance degradation. To address these challenges, we propose a dynamic personalized federated learning framework, called DpFedFKP, for FKP recognition. Specifically, each client first performs local training through classification loss constraints to obtain the local model, and the central server then aggregates these local models via Federated Averaging (FedAvg) to generate a generalized global model. Subsequently, each client performs an adaptive gradient analysis that computes gradients and parameter variations between global and local models using domain-specific sub-datasets. Furthermore, we propose a novel server-side dynamic update mechanism that dynamically adjusts the local model ratios within client-specific personalized models, enabling optimal interpolation between the global and local models to achieve robust generalization and personalization. Comprehensive experiments on two widely used FKP datasets demonstrate that the proposed method has achieved significant performance improvement over the state-of-the-art techniques. Specifically, our method achieves 0.03\% relative accuracy improvement on Data-fkp and 0.52\% relative accuracy improvement on PolyU-fkp, with 12.35\% relative reduction in equal error rate (EER) on PolyU-fkp. Furthermore, the cross-client experiments achieve up to 2.80\% relative accuracy improvement and 33.48\% relative EER reduction. Significantly, the proposed DpFedFKP has strong compatibility with differential privacy techniques, thereby enhancing privacy-preserving capability.",

keywords = "Dynamic update, Federated learning, Finger knuckle print recognition, Generalization and personalization, Privacy-preserving",

author = "Shuyi Li and Jianian Hu and Bob Zhang and Shanping Yu and Lifang Wu",

note = "Publisher Copyright: {\textcopyright} 2026 Elsevier Ltd",

year = "2026",

month = jun,

doi = "10.1016/j.patcog.2025.112966",

language = "English",

volume = "174",

journal = "Pattern Recognition",

issn = "0031-3203",

publisher = "Elsevier Ltd.",

}

TY - JOUR

T1 - DpFedFKP

T2 - Dynamic personalized federated learning for finger knuckle print recognition

AU - Li, Shuyi

AU - Hu, Jianian

AU - Zhang, Bob

AU - Yu, Shanping

AU - Wu, Lifang

PY - 2026/6

Y1 - 2026/6

N2 - Finger knuckle print (FKP) recognition has recently attracted significant attention owing to its low cost, high security, and discriminative characteristics. However, existing data-driven approaches, particularly deep learning methods, typically require centralized training with large-scale datasets, raising serious privacy risks. Furthermore, the substantial data heterogeneity among independent datasets often leads to training difficulties and performance degradation. To address these challenges, we propose a dynamic personalized federated learning framework, called DpFedFKP, for FKP recognition. Specifically, each client first performs local training through classification loss constraints to obtain the local model, and the central server then aggregates these local models via Federated Averaging (FedAvg) to generate a generalized global model. Subsequently, each client performs an adaptive gradient analysis that computes gradients and parameter variations between global and local models using domain-specific sub-datasets. Furthermore, we propose a novel server-side dynamic update mechanism that dynamically adjusts the local model ratios within client-specific personalized models, enabling optimal interpolation between the global and local models to achieve robust generalization and personalization. Comprehensive experiments on two widely used FKP datasets demonstrate that the proposed method has achieved significant performance improvement over the state-of-the-art techniques. Specifically, our method achieves 0.03% relative accuracy improvement on Data-fkp and 0.52% relative accuracy improvement on PolyU-fkp, with 12.35% relative reduction in equal error rate (EER) on PolyU-fkp. Furthermore, the cross-client experiments achieve up to 2.80% relative accuracy improvement and 33.48% relative EER reduction. Significantly, the proposed DpFedFKP has strong compatibility with differential privacy techniques, thereby enhancing privacy-preserving capability.

AB - Finger knuckle print (FKP) recognition has recently attracted significant attention owing to its low cost, high security, and discriminative characteristics. However, existing data-driven approaches, particularly deep learning methods, typically require centralized training with large-scale datasets, raising serious privacy risks. Furthermore, the substantial data heterogeneity among independent datasets often leads to training difficulties and performance degradation. To address these challenges, we propose a dynamic personalized federated learning framework, called DpFedFKP, for FKP recognition. Specifically, each client first performs local training through classification loss constraints to obtain the local model, and the central server then aggregates these local models via Federated Averaging (FedAvg) to generate a generalized global model. Subsequently, each client performs an adaptive gradient analysis that computes gradients and parameter variations between global and local models using domain-specific sub-datasets. Furthermore, we propose a novel server-side dynamic update mechanism that dynamically adjusts the local model ratios within client-specific personalized models, enabling optimal interpolation between the global and local models to achieve robust generalization and personalization. Comprehensive experiments on two widely used FKP datasets demonstrate that the proposed method has achieved significant performance improvement over the state-of-the-art techniques. Specifically, our method achieves 0.03% relative accuracy improvement on Data-fkp and 0.52% relative accuracy improvement on PolyU-fkp, with 12.35% relative reduction in equal error rate (EER) on PolyU-fkp. Furthermore, the cross-client experiments achieve up to 2.80% relative accuracy improvement and 33.48% relative EER reduction. Significantly, the proposed DpFedFKP has strong compatibility with differential privacy techniques, thereby enhancing privacy-preserving capability.

KW - Dynamic update

KW - Federated learning

KW - Finger knuckle print recognition

KW - Generalization and personalization

KW - Privacy-preserving

UR - https://www.scopus.com/pages/publications/105027177205

U2 - 10.1016/j.patcog.2025.112966

DO - 10.1016/j.patcog.2025.112966

M3 - Article

AN - SCOPUS:105027177205

SN - 0031-3203

VL - 174

JO - Pattern Recognition

JF - Pattern Recognition

M1 - 112966

ER -

DpFedFKP: Dynamic personalized federated learning for finger knuckle print recognition

摘要

访问文件

其它文件与链接

学术指纹

引用此