Research on mechanical fault diagnosis method based on federated differential equations DOI
Li Zhi,

B.L. Wang,

Fengtao Wang

et al.

Journal of Vibration and Control, Journal Year: 2025, Volume and Issue: unknown

Published: May 19, 2025

The existing fault diagnosis methods based on federated learning mainly focus the privacy, data heterogeneity, and communication efficiency, while neglecting problem related to memory consumption, computational costs, interpretability in diagnostic model. To overcome these deficiencies, neural networks are re-examined from perspective of dynamic systems this paper. However closely differential equations, problems can usually be described by establishing equations. Therefore, a mechanical equations (FDEs) is proposed. In proposed FDE-based method, complex calculation process between neurons network layers replaced solver, which greatly reduces consumption number model parameters, increases model, establishes connection dynamics learning; deep integration prompted. Finally, FDE method has been successfully applied aero-engine, compared with learning. experimental results show that not only satisfactory recognition rate but also ability continuous parameters reduced, enhanced. research paper important theoretical value engineering application for artificial intelligence.

Language: Английский

Segmentation guided dual-branch classification for measuring fat infiltration in paraspinal muscles DOI
Jing Chen, Hao Jiang, Yiheng Li

et al.

Expert Systems with Applications, Journal Year: 2025, Volume and Issue: unknown, P. 127260 - 127260

Published: March 1, 2025

Language: Английский

Citations

0
CGKDFL: A Federated Learning Approach Based on Client Clustering and Generator‐Based Knowledge Distillation for Heterogeneous Data DOI
S. Zhang,

Hongzhen Xu,

Xiaojun Yu

et al.

Concurrency and Computation Practice and Experience, Journal Year: 2025, Volume and Issue: 37(9-11)

Published: April 8, 2025

ABSTRACT In practical, real‐world complex networks, data distribution is frequently decentralized and Non‐Independently Identically Distributed (Non‐IID). This heterogeneous presents a significant challenge for federated learning. Such problems include the generation of biased global models, lack sufficient personalization capability local difficulty in absorbing knowledge. We propose Federated Learning Approach Based on Client Clustering Generator‐based Knowledge Distillation(CGKDFL) data. Firstly, to reduce model bias, we clustering learning approach that only requires each client transmit some parameters selected layer, thus reducing number parameters. Subsequently, circumvent absence knowledge resulting from clustering, generator designed improve privacy features increase diversity developed server side. produces feature representation aligns with specific tasks by utilizing labeling information provided client. achieved without need any external dataset. The then transfers its model. can utilize this distillation. Finally, extensive experiments were conducted three datasets. results demonstrate CGKDFL outperforms baseline method minimum , regarding accuracy Additionally, it compared methods convergence speed all cases.

Language: Английский

Citations

0
Accelerating Federated Learning with genetic algorithm enhancements DOI

Huanqing Zheng,

Jielei Chu, Zhaoyu Li

et al.

Expert Systems with Applications, Journal Year: 2025, Volume and Issue: 281, P. 127636 - 127636

Published: April 17, 2025

Language: Английский

Citations

0
FedAgent: Federated learning on Non-IID data via reinforcement learning and knowledge distillation DOI
Bingli Sun, Xiao Song, Yuchun Tu

et al.

Expert Systems with Applications, Journal Year: 2025, Volume and Issue: unknown, P. 127973 - 127973

Published: May 1, 2025

Language: Английский

Citations

0
A privacy-enhanced framework for collaborative Big Data analysis in healthcare using adaptive federated learning aggregation DOI Creative Commons

R Haripriya,

Nilay Khare, Manish Pandey

et al.

Journal Of Big Data, Journal Year: 2025, Volume and Issue: 12(1)

Published: May 6, 2025

Language: Английский

Citations

0
Research on mechanical fault diagnosis method based on federated differential equations DOI
Li Zhi,

B.L. Wang,

Fengtao Wang

et al.

Journal of Vibration and Control, Journal Year: 2025, Volume and Issue: unknown

Published: May 19, 2025

The existing fault diagnosis methods based on federated learning mainly focus the privacy, data heterogeneity, and communication efficiency, while neglecting problem related to memory consumption, computational costs, interpretability in diagnostic model. To overcome these deficiencies, neural networks are re-examined from perspective of dynamic systems this paper. However closely differential equations, problems can usually be described by establishing equations. Therefore, a mechanical equations (FDEs) is proposed. In proposed FDE-based method, complex calculation process between neurons network layers replaced solver, which greatly reduces consumption number model parameters, increases model, establishes connection dynamics learning; deep integration prompted. Finally, FDE method has been successfully applied aero-engine, compared with learning. experimental results show that not only satisfactory recognition rate but also ability continuous parameters reduced, enhanced. research paper important theoretical value engineering application for artificial intelligence.

Language: Английский

Citations

0