ABSTRACT
As the perceptual brain decoding, a person's cognitive processes can be detected by their brain activity patterns. EEG signals, being physiological biosignals, are associated with human cortical activities with high temporal resolution. While such a significant advancement has grabbed the interest of a wide spectrum of research communities, EEG-based classification of brain activities elicited by images still needs to be improved in terms of accuracy, generality, and interpretation. While such a great increase has attracted a huge variety of investigations among applicable studies communities, nevertheless needs efforts for similar development with appreciate to its accuracy, generalization, and interpretation.
This chapter shows the use of perceptual brain decoding to classify EEG brain waves using an image classification approach. The proposed method works by representing images and classifying brain waves signals using machine learning and deep learning algorithms. We have used the publicly available EEG data from Mind Big data [1,2]. We prepared an end-to-end pipeline and conclude with our result that CNN architecture yields the best accuracy of 73.54%, with LSTM a close second with 72.91% being computationally easy, and third is DNN architecture with 72.39%.
