A robust framework epileptic seizures classification based on lightweight structure deep convolutional neural network and wavelet decomposition

Nazanin Nemati and Saeed Meshgini and Ali Farzamnia (2020) A robust framework epileptic seizures classification based on lightweight structure deep convolutional neural network and wavelet decomposition. In: 2020 28th Iranian Conference on Electrical Engineering, ICEE 2020, 04 August 2020, University of Tabriz, Iran.

	Text A robust framework epileptic seizures classification based on lightweight structure deep convolutional neural network and wavelet decomposition FULL TEXT.pdf Restricted to Registered users only Download (356kB)
	Text A robust framework epileptic seizures classification based on lightweight structure deep convolutional neural network and wavelet decomposition ABSTRACT.pdf Download (62kB)

Official URL: https://ieeexplore.ieee.org/document/9260797

Abstract

Nowadays scientific evidence suggests that epileptic seizures can appear in the brain signals minutes and even hours prior to their occurrence. Advances in predicting epileptic seizures can promise a robust model in which seizures and irreparable injuries at the time of occurrence can be possible. Most of the previous automated solutions are associated with challenges such as the lack of a proper signal descriptor, the existence of a large number of features and, consequently, the time-consuming analysis, which are not considering the uncertainty issue. In this paper, efficient and fastidious classification is performed by analysing the frequency bands of the input EEG signal via discrete wavelet transform, which is relying on the deep convolutional neural network based classification. Using the EEG signals obtained from the CHEG-MIT Scalp EEG database, the implementation in the desired model is performed and the results show that the proposed model has the best response in detecting the disease from the sample signal and with the highest level of certainty to follow. To solve the uncertainty problem, the repeatability algorithm test is arranged and after K-fold cross-validation, the experimental precision of all the three evaluation factors were equal to 99.34%, 99.53%, and 99.76%, respectively.

Item Type:	Conference or Workshop Item (Paper)
Keyword:	Convolutional neural network , Deep learning , Discrete wavelet decomposition , Epilepsy seizure , Lightweight structure
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering
Department:	FACULTY > Faculty of Engineering
Depositing User:	SAFRUDIN BIN DARUN -
Date Deposited:	31 Jul 2021 16:16
Last Modified:	31 Jul 2021 16:16
URI:	https://eprints.ums.edu.my/id/eprint/28946

Actions (login required)

View Item