An EEG-Based Thought Recognition Using Pseudo-Wigner–Kullback–Leibler Deep Neural Classification

Every human being pursues similar circumstances differently and steers them in a disparate manner. Brain activity based on thought detection plays a dominant role while controlling the state of affairs with electroencephalography (EEG) signals. Hence, a comprehensive and specific analysis of EEG signals is mandatory. Brain activity based on thought detection analysis is essential to comprehend human psychology and mental condition in a specified circumstance. Therefore, developing a more robust method for measuring brain activity and detecting thought with EEG signals is significant. This work proposes the pseudo-Wigner–Kullback–Leibler deep neural classifier (PW-KLDNC) approach for brain activity-based thought detection to make accurate predictions quickly and with low error rates. There are three parts to the PW-KLDNC technique. First, the Gauss–Markov discrete Fourier model is used to pre-process the raw EEG signal. Then, feature extraction is performed using a smooth pseudo-Wigner–Ville model that extracts accurate and relevant features from the computationally effective and noise-minimized EEG signals. We then applied the extracted relevant features to the Kullback–Leibler deep neural classifier for the final classification. On the EEG brainwave dataset, we conducted the experiments and demonstrated the significance of the suggested PW-KLDNC approach comparable to current research studies, wherein we used EEG data to categorize emotional states in subjects. According to experimental results, the proposed technique can speed up emotion recognition for samples of different sizes while maintaining high accuracy and low error rates.