EEG Analysis during Recall of Computational Load Task

This paper describes an analysis of EEG signals during the recall of computational load task (or mental arithmetic). For this purpose, feature values were extracted from measured data from multiple EEG electrodes (F3, F4, FC3, etc.) under two conditions: resting state and mental arithmetic image recall state. When we applied SVM (Support-Vector Machine) using features obtained from brain wave component ratios of 20 subjects, the classification accuracy was 72.8%. Subsequently, to correct the classification accuracy, we focused on the heart rate, which reflects the balance of the autonomic nervous system. As a result, when excluding individuals with relatively fast heart rates from the dataset based on their resting state, the classification accuracy improved with results up to 90.2%. Furthermore, when excluding individuals with relatively slow heart rates from the dataset, the classification accuracy decreased to less than 60%.