Comparison of Machine Learning Algorithms for Recognizing Drowsiness in Drivers using Electroencephalogram (EEG) Signals

Abstract

Drowsiness is one of the major reasons that causes traffic accidents. Thus, its early detection can help preventing accidents by warning the drivers before the unfortunate events. This study focuses on the detection of drowsiness using classification of alpha waves from EEG signals with 25 different machine learning algorithms. The results were evaluated in terms of classification accuracy and classification time. Accordingly, the Bagged Trees and Subspace k-Nearest Neighbor models gave better results in terms of classification accuracy compared to the Tree algorithm methodology, although the classification times are relatively high. Tree Algorithms approach displays optimal features as it serves as both a considerably satisfactory classification accuracy in much shorter times. The requirements in terms of accuracy and time for the recognition of drowsiness should determine the method to be applied.