Dual Contrastive Learning for Self-Supervised ECG Mapping to Emotions and Glucose Levels

Noy Lalzary, Lior Wolf

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We learn to map the ECG signal to emotional or physical states using a 1D-CNN followed by a Transformer. To overcome the limited number of samples, we propose a new self-supervised learning scheme that combines latent space masking with both a temporal prediction task and a matching task, both learned via a contrastive loss. Our method outperforms the existing self-supervised approaches for ECG by a wide margin.

Original languageEnglish
Title of host publication2023 IEEE SENSORS, SENSORS 2023 - Conference Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9798350303872
DOIs
StatePublished - 2023
Event2023 IEEE SENSORS, SENSORS 2023 - Vienna, Austria
Duration: 29 Oct 20231 Nov 2023

Publication series

NameProceedings of IEEE Sensors
ISSN (Print)1930-0395
ISSN (Electronic)2168-9229

Conference

Conference2023 IEEE SENSORS, SENSORS 2023
Country/TerritoryAustria
CityVienna
Period29/10/231/11/23

Keywords

  • ECG
  • Self-Supervised Learning
  • Transformers

Fingerprint

Dive into the research topics of 'Dual Contrastive Learning for Self-Supervised ECG Mapping to Emotions and Glucose Levels'. Together they form a unique fingerprint.

Cite this