Article: Multi-head batch attention-enabled dual learning-based deep spiking convolutional neural network model for epileptic seizure detection Journal: International Journal of Biomedical Engineering and Technology (IJBET) 2026 Vol.51 No.2 pp.175 - 213 Abstract: Epileptic seizure is a transient clinical event resulting from the excessive or synchronous discharge of brain neurons that causes alterations in behaviour. Various approaches have been developed to detect such a condition, yet the models faced significant drawbacks. Hence, to address such drawbacks, the research proposes a multi-head batch attention-enabled dual learning-based deep spiking convolutional neural network (MD2SCN). In which the spiking convolutional neural network (SCNN) is used to effectively capture spatio-temporal features from electroencephalography signals while drastically lowering computational costs compared to traditional methods. Additionally, the multi-head attention (MHA) enables the model to better analyse electroencephalography signals. Experimental results demonstrate a high accuracy of 98.34%, precision of 97.45%, and a recall of 99.31%, while reducing the error rate to 1.66% utilising the CHB-MIT Scalp EEG Dataset in comparison with the existing frameworks. Inderscience Publishers - linking academia, business and industry through research

Title: Multi-head batch attention-enabled dual learning-based deep spiking convolutional neural network model for epileptic seizure detection

Authors: Pankaj Kunekar; Puja Cholke; Rajnikant B. Wagh; Jayshri Sonawane; Vishal Thakare; Atul S. Chaudhari

Addresses: Department of Information Technology, Vishwakarma Institute of Technology, Pune, Maharashtra, India ' Department of Information Technology, Vishwakarma Institute of Technology, Pune, Maharashtra, India ' Department of Computer Engineering, R.C. Patel Institute of Technology, Shirpur, Maharashtra, India ' Department of Computer Engineering, R.C. Patel Institute of Technology, Shirpur, Maharashtra, India ' Department of Computer Engineering, R.C. Patel Institute of Technology, Shirpur, Maharashtra, India ' Department of Computer Engineering, R.C. Patel Institute of Technology, Shirpur, Maharashtra, India

Abstract: Epileptic seizure is a transient clinical event resulting from the excessive or synchronous discharge of brain neurons that causes alterations in behaviour. Various approaches have been developed to detect such a condition, yet the models faced significant drawbacks. Hence, to address such drawbacks, the research proposes a multi-head batch attention-enabled dual learning-based deep spiking convolutional neural network (MD2SCN). In which the spiking convolutional neural network (SCNN) is used to effectively capture spatio-temporal features from electroencephalography signals while drastically lowering computational costs compared to traditional methods. Additionally, the multi-head attention (MHA) enables the model to better analyse electroencephalography signals. Experimental results demonstrate a high accuracy of 98.34%, precision of 97.45%, and a recall of 99.31%, while reducing the error rate to 1.66% utilising the CHB-MIT Scalp EEG Dataset in comparison with the existing frameworks.

Keywords: epileptic seizure detection; electroencephalography; dual learning; multi-head batch attention; spiking network.

DOI: 10.1504/IJBET.2026.156074

International Journal of Biomedical Engineering and Technology, 2026 Vol.51 No.2, pp.175 - 213

Received: 23 Dec 2025
Accepted: 13 Apr 2026
Published online: 02 Sep 2026 *

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

Title: Multi-head batch attention-enabled dual learning-based deep spiking convolutional neural network model for epileptic seizure detection

Keep up-to-date