Brain-Computer Interface: A Novel EEG Classification for Baseline Eye States Using LGBM Algorithm

An enhanced EEG prediction system for motor cortex-imagery tasks using SVM

“…This work used the feature selection step to determine the best channels that have acquired the changes in brain activity during each class with good precision, to properly select these channels, this work uses the Extra Trees algorithm which is one of the best EEG data classifiers [3], knowing that this algorithm allows building a classification tree based on training data and then calculate the degree of importance D of each channel i in using the following relationship:…”

“…Traditional artificial neural networks that are based on the lowest empirical risk outperform the SVM since it is based on the smallest structural risk. This classifier's purpose is to find the best hyperplane for distinguishing each mode class [1,2,3]. The SVM chooses hyperplanes that group the most points of the same class together while keeping the gap between each class and those hyperplanes as little as possible.…”

“…The support vector is created using the hyperplane's nearest point. The shortest path between them and class points is the distance from the class to the hyperplane [1,2,3]. This is how far something is measured.…”

“…The performance of detection and prediction is measured using the metrics as illustrated in Table 1: The most widely used method for comparing algorithm performance in classification without class concentration. As a result, prediction accuracy (AC) is the most frequent means of experiencing to avoid separating the number of excellent labels between distinct tasks [2,3]:…”