Attention-Based CNN for Efficient Embedded ECG Signal Classification

DOI

10.2991/978-94-6463-720-5_14How to use a DOI?

Keywords

ECG classification; attention-based CNN; embedded systems; quantization; default quantization; Float16 quantization; dynamic range quantization; wearable health monitoring

Abstract

This study presents an efficient attention-based convolutional neural network (CNN) for ECG signal classification, tailored for embedded systems. The proposed model combines deep learning with attention mechanisms to enhance accuracy and interpretability. Quantization techniques, including default quantization (weights reduced to 8-bit integers), Float16 quantization (weights reduced to 16-bit floats), and dynamic range quantization (offering the best performance with minimal accuracy loss), were employed to optimize the model for resource-constrained devices. Testing on the ECG Fragment Database for Dangerous Arrhythmia from PhysioNet demonstrated over 97.3% classification accuracy, with the quantized model achieving inference times under 1 second on a Raspberry Pi. These findings showcase the potential for real-time, portable ECG analysis in medical applications.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Madusha Suraweera
AU  - Mahima Weerasinghe
PY  - 2025
DA  - 2025/06/30
TI  - Attention-Based CNN for Efficient Embedded ECG Signal Classification
BT  - Proceedings of the Smart Sustainable Development Conference 2025 (SSD 2025)
PB  - Atlantis Press
SP  - 160
EP  - 170
SN  - 3005-155X
UR  - https://doi.org/10.2991/978-94-6463-720-5_14
DO  - 10.2991/978-94-6463-720-5_14
ID  - Suraweera2025
ER  -