Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ECG-guided individual identification via PPG

Riling Wei, Hanjie Chen, Kelu Yao, Chuanguang Yang, Jun Wang, Chao Li

TL;DR

PPG-based biometric recognition suffers from limited information density; this work addresses the gap by transferring discriminative knowledge from ECG to PPG via a cross-modal KD framework. A plug-and-play CLIP-based knowledge alignment maps both modalities into a shared latent space, and a Cross-Knowledge Assessment module monitors teaching effectiveness, with an overall loss that combines task, KD, relation, and cross-KD terms. Experiments on MIMIC data show state-of-the-art performance in both sample-wise and subject-wise settings, with up to around 3 percentage-point gains in seen/unseen accuracy and reduced EER, demonstrating effective cross-modal knowledge transfer without increasing inference burden. The approach holds promise for robust, non-invasive biometric identification in real-world security applications, though it reports some performance degradation at higher N-shot levels that warrant future investigation.

Abstract

Photoplethsmography (PPG)-based individual identification aiming at recognizing humans via intrinsic cardiovascular activities has raised extensive attention due to its high security and resistance to mimicry. However, this kind of technology witnesses unpromising results due to the limitation of low information density. To this end, electrocardiogram (ECG) signals have been introduced as a novel modality to enhance the density of input information. Specifically, a novel cross-modal knowledge distillation framework is implemented to propagate discriminate knowledge from ECG modality to PPG modality without incurring additional computational demands at the inference phase. Furthermore, to ensure efficient knowledge propagation, Contrastive Language-Image Pre-training (CLIP)-based knowledge alignment and cross-knowledge assessment modules are proposed respectively. Comprehensive experiments are conducted and results show our framework outperforms the baseline model with the improvement of 2.8% and 3.0% in terms of overall accuracy on seen- and unseen individual recognitions.

ECG-guided individual identification via PPG

TL;DR

PPG-based biometric recognition suffers from limited information density; this work addresses the gap by transferring discriminative knowledge from ECG to PPG via a cross-modal KD framework. A plug-and-play CLIP-based knowledge alignment maps both modalities into a shared latent space, and a Cross-Knowledge Assessment module monitors teaching effectiveness, with an overall loss that combines task, KD, relation, and cross-KD terms. Experiments on MIMIC data show state-of-the-art performance in both sample-wise and subject-wise settings, with up to around 3 percentage-point gains in seen/unseen accuracy and reduced EER, demonstrating effective cross-modal knowledge transfer without increasing inference burden. The approach holds promise for robust, non-invasive biometric identification in real-world security applications, though it reports some performance degradation at higher N-shot levels that warrant future investigation.

Abstract

Photoplethsmography (PPG)-based individual identification aiming at recognizing humans via intrinsic cardiovascular activities has raised extensive attention due to its high security and resistance to mimicry. However, this kind of technology witnesses unpromising results due to the limitation of low information density. To this end, electrocardiogram (ECG) signals have been introduced as a novel modality to enhance the density of input information. Specifically, a novel cross-modal knowledge distillation framework is implemented to propagate discriminate knowledge from ECG modality to PPG modality without incurring additional computational demands at the inference phase. Furthermore, to ensure efficient knowledge propagation, Contrastive Language-Image Pre-training (CLIP)-based knowledge alignment and cross-knowledge assessment modules are proposed respectively. Comprehensive experiments are conducted and results show our framework outperforms the baseline model with the improvement of 2.8% and 3.0% in terms of overall accuracy on seen- and unseen individual recognitions.
Paper Structure (15 sections, 9 equations, 4 figures, 4 tables)

This paper contains 15 sections, 9 equations, 4 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Methodology
  4. CLIP-based Knowledge Alignment (CLIP)
  5. Cross-Knowledge Assessment (CKA)
  6. Experimental Setup
  7. Dataset
  8. Experiments and Evaluation Metrics
  9. Results and Discussion
  10. Sample-wise Testing
  11. Subject-wise Testing
  12. Compare with SOTA methods
  13. Ablation studies
  14. Visualization
  15. Conclusion

Figures (4)

  • Figure 1: Visualizing the distribution of ECG modalities (the orange part) and PPG modalities (the blue part).
  • Figure 2: The training stage of CLIP-based knowledge alignment module.
  • Figure 3: The framework of the proposed methods.
  • Figure 4: The visualization of feature distribution in sample-wise testing. The red and grey parts denote student and teacher representations respectively. (1) shows that representations of the baseline. (2) indicates the feature distributions of the proposed framework.