EEG Signal-Based Eye Blink Classifier using Random Forest for BCI Systems

The objective of the present work is to perform a feature extraction and classification of three eye blinking scenarios: eyes open, eyes closed and voluntary blinks, to be later used in a BCI (Brain-Computer Interface). Electroencephalographic signals from 6 healthy subjects (3 men and 3 women) between 21 and 28 years have been recorded. The extracted features were the variance and the bandpower. The signals were analyzed with MATLAB’s help and then Random Forest (RF) and Support Vector Machine (SVM) classification methods from the Weka software were used. A comparison among the extracted features has been performed in order to observe which of them are better for classification. According to the results, the variance has an accuracy of 80.7%; bandpower, 33.3%; and a combination of both, 78.1%, by using the RF classifier.