USE OF ARTIFICIAL INTELLIGENCE IN BIOMETRIC AUTHENTICATION SYSTEMS BASED ON DYNAMIC SIGNATURE

A. S. Sverstyuk, H. B. Humeniuk, O. S. Voloshyn

Abstract


The article explores contemporary approaches to applying artificial intelligence (AI) technologies in biometric authentication systems based on dynamic signatures. A scientometric analysis of research trends in this field was conducted using data from the Scopus database covering the period from 2004 to 2025. The results reveal a steady increase in scientific interest in employing AI methods for behavioral biometric authentication, particularly after 2020. Special emphasis is placed on deep learning techniques, including convolutional neural networks (CNNs) and recurrent neural networks (RNNs), which significantly enhance the accuracy, reliability, and speed of signature verification systems. The study also identifies leading research institutions, authors, and publication sources contributing to advancements in biometric technologies. A classification of scientific publications and the primary subject areas where biometric authentication is applied was performed. The findings indicate that hybrid neural network architectures, especially CNN-LSTM models, demonstrate superior efficiency in dynamic signature recognition, outperforming classical algorithms by approximately 10-15% in accuracy. The paper discusses the role of dynamic signatures as a behavioral biometric characteristic that reflects individual motor and psychological features of handwriting. Integrating AI with biometric authentication systems facilitates the development of adaptive, personalized, and more secure identity verification mechanisms applicable in banking, healthcare, military, and educational information systems. Simultaneously, the study highlights several challenges associated with using AI in biometric authentication, including personal data protection, potential falsification of digital biometric templates, and ethical concerns related to automated decision-making systems. Prospective research directions are proposed, including the development of multibiometric authentication systems, adaptive model training techniques, and the integration of dynamic signature verification with blockchain technologies to enhance security and data integrity.
The study also examines the biological basis of dynamic signature formation, emphasizing the role of neurophysiological and motor processes in shaping individual handwriting patterns. It demonstrates that dynamic signature characteristics are determined by the functioning of the central nervous system, motor coordination, and muscle memory, which together ensure the uniqueness of each individual’s signature. Incorporating biological and behavioral features into artificial intelligence models enhances the accuracy, adaptability, and reliability of biometric authentication systems.

Keywords


artificial intelligence; biometric authentication; dynamic signature; deep learning; neural networks; CNN; RNN; information security; blockchain

References


Martseniuk V. P., Sverstiuk A. S., Bahriy Zaiats O. A., Nykytiuk S. O., Prodan A. M. Naukovyi analitychnyi poshuk doslidzhen iz vykorystannia biosensoriv u hromadskomu zdorovi. Visnyk sotsialnoi hihiieny ta orhanizatsii okhorony zdorovia Ukrainy. 2025. T. 96, No 3. S. 183–189. https://doi.org/10.11603/1681 2786.2025.3.15667 [in Ukrainian]

Panychev V. O., Sverstiuk A. S., Kuchvara O. M., Bahriy Zaiats O. A., Maykhruk Z. V. Naukovyi analitychnyi poshuk doslidzhennia problem poshyrennia hostrykh respiratornykh zakhvoriuvan, hrypu ta COVID 19. Visnyk sotsialnoi hihiieny ta orhanizatsii okhorony zdorovia Ukrainy. 2024. T. 92, No 3. S. 76–83. https://doi.org/10.11603/1681-2786.2024.3.14950 [in Ukrainian]

Çiftçi B., Tekin R. Deep learning based offline handwritten signature recognition. Bitlis Eren Univ. J. Sci. 2024. Vol. 13, № 3. P. 871–884. https://doi.org/10.31591/buefad.13619.

Danna J., Velay J. L. Basic and supplementary sensory feedback in handwriting. Frontiers in Psychology. 2015. Vol. 6. Article 169. https://doi.org/10.3389/fpsyg.2015.00169.

Ferrer M. A., Diaz M., Impedovo D., Pirlo G. Offline signature verification: Current state of the art and future trends. Pattern Recognition. 2021. Vol. 111. P. 107731. https://doi.org/10.1016/j.patcog.2020.107731.

Hafemann L. G., Sabourin R., Oliveira L. S. Writer independent feature learning for offline signature verification using deep convolutional neural networks. Pattern Recognition. 2020. Vol. 103. P. 107321. https://doi.org/10.1016/j.patcog.2020.107321.

Impedovo D., Pirlo G. Dynamic handwriting analysis for the assessment of neurodegenerative diseases: A pattern recognition perspective. IEEE Reviews in Biomedical Engineering. 2019. Vol. 12. P. 209–220. https://doi.org/10.1109/RBME.2018.2875019.

Jatav M., Soni S. CNN BiLSTM architectures for handwritten signature verification: Insights and innovations. Journal of Image Processing & Pattern Recognition Progress. 2025. Vol. 12, № 2. P. 43–50. https://doi.org/10.1234/jippr.2025.12.2.43.

Kumar R., Sharma R. Deep learning approaches for dynamic signature verification: A review. IEEE Access. 2021. Vol. 9. P. 130123–130145. https://doi.org/10.1109/ACCESS.2021.3094136.

Parziale A., Diaz M., Ferrer M. A., Marcelli A. Stability modulated dynamic time warping for signature verification. Pattern Recognition Letters. 2024, Vol. 178. P. 12–18. https://doi.org/10.1016/j.patrec.2024.03.009.

Plamondon R., Srihari S. N. Online and offline handwriting recognition: A comprehensive survey. IEEE Transactions on Pattern Analysis and Machine Intelligence. 2000. Vol. 22, № 1. P. 63–84. https://doi.org/10.1109/34.824821.

Rosenblum S., Weiss P. L., Parush S. Product and process evaluation of handwriting difficulties. Educational Psychology Review. 2003. Vol. 15. P. 41–81. https://doi.org/10.1023/A:1021371425220.

Van Galen G. P. Motor control of handwriting: I. Theoretical framework. Acta Psychologica. 1991. Vol. 77. P. 69–88. https://doi.org/10.1016/0001-6918(91)90004-G.




DOI: https://doi.org/10.25128/2078-2357.25.4.7

Refbacks

  • There are currently no refbacks.


Creative Commons Attribution 4.0 License