Open AccessUsing neural network for drowsiness detection based on EEG signals and optimization in the selection of its features using genetic algorithm
Author(s) -
Sepehr Sarabi,
Milad Asadnejad,
Saman Rajabi
Publication year - 2020
Publication title -
innovaciencia
Language(s) - English
Resource type - Journals
ISSN - 2346-075X
DOI - 10.15649/2346075x.1004
Subject(s) - artificial neural network , computer science , electroencephalography , artificial intelligence , genetic algorithm , pattern recognition (psychology) , linear discriminant analysis , cluster analysis , selection (genetic algorithm) , population , perceptron , algorithm , machine learning , psychology , demography , psychiatry , sociology
One of the major causes of traffic accidents is driver’s drowsiness. For this reason, detecting whether the driver's eyes are open or closed is one of the critical factors in reducing road deaths. One way to detect whether your eyes are open or closed is to use EEG signals. EEG signals are obtained from the recording of electrical activity in the human brain. The present study uses a neural network that is applied to the driver's EEG signals to detect whether the eye is open or closed. The data of the EEG signals used in this paper consist of 14 features that are based on a statistical population of 600 people. Various neural network algorithms have been implemented for clustering these data into two classes of open or closed eyes, which are described in this paper. Perceptron neural network and radial base neural network (RBF) are two types of networks used in this paper. Also, in order to improve the execution speed and reduce the occupied space of the microcontroller, the genetic algorithm method has been used to optimize the fitting function of Fisher’s discriminant rate, in which the optimized function provides better results in the less occupied time and space.