Hardware-oriented pruning and quantization of Deep Learning models to detect life-threatening arrhythmias

Lizeth Gonzalez-Carabarin, Alexandre Schmid, Ruud J.G. van Sloun

Onderzoeksoutput: Hoofdstuk in Boek/Rapport/CongresprocedureConferentiebijdrageAcademicpeer review

3 Citaten (Scopus)

Samenvatting

Wearable solutions based on Deep Learning (DL) for real-time ECG monitoring are a promising alternative to detect life-threatening arrhythmias. However, DL models suffer of a large memory footprint, which hampers their adoption in portable technologies. Therefore, we leverage a hardware-oriented pruning approach to effectively shrink DL models. We demonstrate that tiny DL models can be reduced to 5.55x (pruning), and 26.6x (pruning+quantization) compression rate, with 82.9% FLOP's reduction. These ultra-compressed models are able to effectively classify life-threatening arrhythmias with minimal or no loss of performance compared with their non-pruned counterparts, which can pave the path towards DL-based biomedical portable solutions.

Originele taal-2Engels
Titel2021 IEEE Biomedical Circuits and Systems Conference (BioCAS)
UitgeverijInstitute of Electrical and Electronics Engineers
Aantal pagina's6
ISBN van elektronische versie978-1-7281-7204-0
DOI's
StatusGepubliceerd - 23 dec. 2021
Evenement2021 IEEE Biomedical Circuits and Systems Conference, BioCAS 2021 - Virtual, Online, Berlin, Virtual, Online, Duitsland
Duur: 6 okt. 20219 okt. 2021
https://2021.ieee-biocas.org/

Congres

Congres2021 IEEE Biomedical Circuits and Systems Conference, BioCAS 2021
Verkorte titelBioCAS
Land/RegioDuitsland
StadVirtual, Online
Periode6/10/219/10/21
Internet adres

Bibliografische nota

Publisher Copyright:
© 2021 IEEE.

Financiering

This work is funded by Marie Sklodowska - Curie Actions

Vingerafdruk

Duik in de onderzoeksthema's van 'Hardware-oriented pruning and quantization of Deep Learning models to detect life-threatening arrhythmias'. Samen vormen ze een unieke vingerafdruk.

Citeer dit