DefinePK

Abstract

This paper aims to present the Convolutional Neural Network (CNN) model to differentiate attention from non-attention conditions using spontaneous electroencephalogram (EEG) signals. The CNN model was constructed to acquire a general concept to classify attention conditions. A total of 30 subjects were recruited voluntarily for the data acquisition purpose. The experimental performance was benchmarked with the commonly used non-convolution learning algorithms, the support vector machine (SVM). The coherence feature extraction method was used to generate the training data for non-convolution model. The experimental results show that the proposed CNN model has accurately classify 63.89% of the test cases. It has outperformed the SVM model with 4.45% of improvement. In summary, the CNN model is able to create a decent attention classification model using spontaneous EEG signals.