Title: SiamEEGNet: few-shot learning for electroencephalogram-based biometric recognition system

Authors: Kriti Srivastava; Siddharth Sanghavi; Parag Vaid; Palash Rathod

Addresses: Department of Computer Science and Engineering (Data Science), Dwarkadas J. Sanghvi College of Engineering, Mumbai, Maharashtra, 400056, India ' Department of Computer Science and Engineering (Data Science), Dwarkadas J. Sanghvi College of Engineering, Mumbai, Maharashtra, 400056, India ' Department of Computer Science and Engineering (Data Science), Dwarkadas J. Sanghvi College of Engineering, Mumbai, Maharashtra, 400056, India ' Department of Computer Science and Engineering (Data Science), Dwarkadas J. Sanghvi College of Engineering, Mumbai, Maharashtra, 400056, India

Abstract: Authentication is verifying a user's identity when they enter a system. Due to their distinctiveness, biometric-based authentication solutions have started displacing conventional systems. This study suggests employing Electroencephalogram (EEG) or brain waves as a biometric modality since the level of uniqueness attained is higher. These noise-free ECG beats generate greyscale images using the proposed SiamEEGNet. A customised activation mechanism is also designed in this study to hasten the integration of the SiamEEGNet. The one that is suggested can extract characteristics using provided data. EEG signals are difficult to manually analyse and extract features from since they are highly dimensional and have a low signal-to-noise ratio. Because deep learning architectures have transformed end-to-end learning, this study suggests employing them. Convolutional Siamese Neural Networks are used by the suggested method, SiamEEGNet, to perform few-shot learning on a well-known and openly accessible dataset called EEG Motor Movement/Imagery Dataset (EEG- MMIDB), which consists of 106 subjects. The model is then quantitatively assessed using several criteria for person identification and authentication. SiamEEGNet competes favourably with current cutting-edge methods.

Keywords: biometric recognition; Siamese neural networks; CNNs; convolutional neural networks; PhysioNet; electroencephalogram; spectrogram.

DOI: 10.1504/IJSSE.2025.148728

International Journal of System of Systems Engineering, 2025 Vol.15 No.4, pp.320 - 344

Received: 30 May 2023
Accepted: 23 Jul 2023
Published online: 22 Sep 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article