Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Decoding Listeners Identity: Person Identification from EEG Signals Using a Lightweight Spiking Transformer

Zheyuan Lin, Siqi Cai, Haizhou Li

TL;DR

This work tackles EEG-based person identification under energy constraints by introducing a lightweight spiking transformer that fuses Conv-based SNN blocks with Spike-driven Self-Attention. The architecture leverages Leaky Integrate-and-Fire neurons and a hybrid block design to efficiently process EEG windows, achieving $3.91\mathrm{M}$ parameters and $760.7\ \mu\text{J}$ per forward pass while maintaining high accuracy. On the EEG-MER Challenge dataset with 26 subjects, the method attains 100% test accuracy and demonstrates substantial energy savings compared to non-SNN baselines, highlighting its suitability for real-time, edge-based BCIs. These results indicate a promising direction for robust, low-power biometric EEG systems applicable to personalized and secure brain-computer interfaces.

Abstract

EEG-based person identification enables applications in security, personalized brain-computer interfaces (BCIs), and cognitive monitoring. However, existing techniques often rely on deep learning architectures at high computational cost, limiting their scope of applications. In this study, we propose a novel EEG person identification approach using spiking neural networks (SNNs) with a lightweight spiking transformer for efficiency and effectiveness. The proposed SNN model is capable of handling the temporal complexities inherent in EEG signals. On the EEG-Music Emotion Recognition Challenge dataset, the proposed model achieves 100% classification accuracy with less than 10% energy consumption of traditional deep neural networks. This study offers a promising direction for energy-efficient and high-performance BCIs. The source code is available at https://github.com/PatrickZLin/Decode-ListenerIdentity.

Decoding Listeners Identity: Person Identification from EEG Signals Using a Lightweight Spiking Transformer

TL;DR

This work tackles EEG-based person identification under energy constraints by introducing a lightweight spiking transformer that fuses Conv-based SNN blocks with Spike-driven Self-Attention. The architecture leverages Leaky Integrate-and-Fire neurons and a hybrid block design to efficiently process EEG windows, achieving parameters and per forward pass while maintaining high accuracy. On the EEG-MER Challenge dataset with 26 subjects, the method attains 100% test accuracy and demonstrates substantial energy savings compared to non-SNN baselines, highlighting its suitability for real-time, edge-based BCIs. These results indicate a promising direction for robust, low-power biometric EEG systems applicable to personalized and secure brain-computer interfaces.

Abstract

EEG-based person identification enables applications in security, personalized brain-computer interfaces (BCIs), and cognitive monitoring. However, existing techniques often rely on deep learning architectures at high computational cost, limiting their scope of applications. In this study, we propose a novel EEG person identification approach using spiking neural networks (SNNs) with a lightweight spiking transformer for efficiency and effectiveness. The proposed SNN model is capable of handling the temporal complexities inherent in EEG signals. On the EEG-Music Emotion Recognition Challenge dataset, the proposed model achieves 100% classification accuracy with less than 10% energy consumption of traditional deep neural networks. This study offers a promising direction for energy-efficient and high-performance BCIs. The source code is available at https://github.com/PatrickZLin/Decode-ListenerIdentity.
Paper Structure (18 sections, 7 equations, 2 figures, 1 table)

This paper contains 18 sections, 7 equations, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. Methodology
  3. Spiking Neurons
  4. Model Architecture
  5. Overall Model
  6. Conv-based SNN block
  7. Transformer-based SNN block
  8. Parameter Reduction
  9. Dataset and Experiments
  10. Dataset
  11. Data Preprocessing
  12. Training Protocol
  13. Results
  14. Person Identification Accuracy
  15. Model Efficiency
  16. ...and 3 more sections

Figures (2)

  • Figure 1: Architecture overview of the proposed lightweight spiking transformer for EEG-based person identification. It takes a window of EEG signals as input, and performs multi-class classification to determine the individual's identity. $\bigoplus$ stands for shortcuts.
  • Figure 2: Outline of the Spike-Driven Self-Attention (SDSA) block, RepConv stands for re-parameterization convolution.