Tianyi Zhang¹, Fengzhi Dai^1,2, Di Yin¹, Jichao Zhao¹
1Tianjin University of Science and Technology, China
2Tianjin Tianke Intelligent and Manufacture Technology CO., LTD, China
pp. 1–5
ABSTRACT
Research shows that human emotion is closely related to the activity correlation of cerebral cortex, so the research of emotion classification by EEG (Electroencephalogram) provides a reliable basis. The feature extraction and classification application for EEG has been greatly improved in recent years, so we use EEG to study emotion classification. However, there are differences between EEG signals of different subjects, which have a certain impact on emotion classification. How to ensure the high accuracy and robustness of recognition is a problem. For this problem, when studying different subjects in different states, spectrum analysis can be used for their feature extraction. When the extracted features are classified, discriminant analysis algorithm is used and achieved better classification results. There are many methods involved in feature extraction, and different feature extraction methods will be compared later, so as to improve the robustness and efficiency of emotional classification by EEG signals.

ARTICLE INFO
Article History
Received 25 October 2019
Accepted 28 February 2021

Keywords
EEG
Feature extraction
Channel selection
Spectrum analysis
Sentiment classification

JAALR2101

Download article(PDF)