Efficient and secure µ -Training and µ -Fine-Tuning for edge-based TinyML with future-guided self-distillation
Machine Learning: Health · 2025
DOI: 10.1088/3049-477x/add26a ↗Kurz gefasst
Fachartikel in der Zeitschrift Machine Learning: Health, 2025. Verfasst von Zhaojing Huang und 3 weiteren Autor:innen. Die Arbeit nennt 43 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
This study presents a novel, computationally efficient training framework demonstrated through bio-signal processing on edge medical devices. The approach integrates conventional full training with an innovative µ -Training technique, wherein the encoder and decoder of a compact model remain frozen while only the middle layer is updated. This design is further enhanced by a novel Future-Guided Self-Distillation mechanism that leverages the model’s anticipated future state in training to boost performance and improve generalization on unseen data, using electrocardiogram signals as the primary case study. Additionally, µ -Fine-Tuning facilitates on-device adaptation under resource-constrained conditions. We validate our framework using in-sample data from the Telehealth Network of Minas Gerais and out-of-sample testing on the China Physiological Signal Challenge 2018 datasets. Experimental results demonstrate that our integrated strategy (combining full training, self-distilled µ -Training, and µ -Fine-Tuning) consistently matches or surpasses conventional methods while significantly improving computational efficiency and mitigating catastrophic forgetting. Deployment on Radxa Zero hardware underscores the approach’s practical applicability and scalability. Moreover, a demonstration incorporating the proposed self-distilled µ -Training into standard training procedures reveals performance improvements. This highlights the technique’s potential for broader applications beyond medical diagnostics and tiny machine learning systems, paving the way for its integration into existing training mechanisms to elevate overall model performance.
Zitieren
APA
Huang, Z., Yu, L., Contreras, L. F. H., & Kavehei, O. (2025). Efficient and secure µ -Training and µ -Fine-Tuning for edge-based TinyML with future-guided self-distillation. Machine Learning: Health. https://doi.org/10.1088/3049-477x/add26a
Vancouver
Huang Z, Yu L, Contreras LFH, Kavehei O. Efficient and secure µ -Training and µ -Fine-Tuning for edge-based TinyML with future-guided self-distillation. Machine Learning: Health. 2025. doi:10.1088/3049-477x/add26a
BibTeX
@article{Huang2025,
author = {Huang, Zhaojing and Yu, Leping and Contreras, Luis Fernando Herbozo and Kavehei, Omid},
title = {Efficient and secure µ -Training and µ -Fine-Tuning for edge-based TinyML with future-guided self-distillation},
journal = {Machine Learning: Health},
year = {2025},
doi = {10.1088/3049-477x/add26a}
}
EUVIMED