A Novel Machine Learning-Based Analysis of Affects using EEG Data

The main topic of study is mapping human cognition into automated analysis since it has exciting applications in practically every aspect of creating artificially intelligent devices. Studying electroencephalogram (EEG) patterns is the greatest approach to comprehending how the brain functions; hence a lot of research has been focused on this topic. It is challenging to create a generic affect classification system that can effectively deliver robust affect labeling to the EEG patterns since EEG recordings are subject-dependent and show variances according to external influences or kinds of recording devices. A unique generic framework for affect-based cognitive analysis is presented in the proposed study as a solution to this problem. The proposed system includes the following steps: EEG pattern, dataset, pre-processing, feature selection, an Enhanced Squirrel search optimized Naive Bayes (ESSOp-NB) step to reduce inter-class and intra-class variance, and finally, the processed pattern is sent to a trained classifier for classification into the proper effect categories. The effectiveness of well-known classifiers is evaluated using EEG data from single and multiple people from two separate datasets: Database for Emotion Analysis using Physiological Signals (DEAP) and SJTU Emotion EEG Dataset (SEED). The findings of the experiment indicate that the ESSOp-NB classifier is the most effective at categorizing data from both single and mixed participants. &nbsp