Cognitive load detection using Binary salp swarm algorithm for feature selection
Jammisetty Yedukondalu, Lakhan Dev Sharma
Abstract
Analyzing the brain's reaction to stimuli requires the detection of cognitive load during the mental assignment of neuronal activity. It is possible to determine the cognitive load experienced during mental arithmetic tasks using an electroencephalogram (EEG). Data from the mental arithmetic task (MAT) was gathered using EEG. The EEG signals were decomposed into intrinsic mode functions (IMFs) using circulant singular spectrum analysis (CiSSA). After that, we extract the entropy-based features from IMFs. The Binary salp swarm algorithm (BSSA) was used to obtain the optimal feature subset for feature selection. Moreover, K-nearest neighbour (KNN) and support vector machine (SVM) machine learning models were used to classify a set of features based on performance indicators including accuracy (AC%), sensitivity (SE), specificity (SP), precision (PR), and F-Score (F-S). The best classification accuracy is achieved by the presented technique, which is 95.28% for KNN classifiers and 94.54% for SVM classifiers, respectively. The results of the study demonstrated that, in comparison to current techniques, the suggested approach is better accurate in detecting cognitive load.