Application and Development of Machine Learning Based Brain Computer Interface Technology in Emotion Recognition

Accompanied with advances in brain computer interface (BCI) technology, recognizing emotion through brain and electroencephalography (EEG) signals is becoming more and more popular and accurate. Models and algorithms to interpret brain signals are explored with different results, aiming at more accurately recognizing and in return effectively regulating emotions. In order to comprehend contemporary emotion recognition (ER) theories based on BCI, EEG, and deep learning, and this paper will review several commonly accepted emotion models and the experimental practice using EEG signals to measure various emotions. In the following, approaches to decipher EEG signals, including feature extraction in different domains, electrode selection, classification of emotions, and brain network, will be explained and compared in accuracy. The paper will also discuss future research direction, proposed application, as well as challenges of BCI-based ER. The results of review show a noticeable emergence in novel ER algorithms and increase in reliability and accuracy. This also reflects a higher and more meaningful application in future ER practices, possibly, and in large scale, improving people’s living status, helping them understand themselves, regulating violence in society, and avoiding negative impacts of depression and related emotion pressure.