EEG-Based emotion recognition in listening music by using support vector machine and linear dynamic system

Authors:
Ruo-Nan Duan;Xiao-Wei Wang;Bao-Liang Lu
Affiliations:
Center for Brain-Like Computing and Machine Intelligence, Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai, China;Center for Brain-Like Computing and Machine Intelligence, Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai, China;Dept. of Comp. Science and Eng., Shanghai Jiao Tong Univ., China, MOE-Microsoft Key Lab. for Intelligent Computing and Intelligent Systems, Shanghai Jiao Tong Univ., China, Shanghai Key Lab. of Sc ...
Venue:
ICONIP'12 Proceedings of the 19th international conference on Neural Information Processing - Volume Part IV
Year:
2012

Citing 4
Cited 0

Feature Selection Based on Mutual Information: Criteria of Max-Dependency, Max-Relevance, and Min-Redundancy

IEEE Transactions on Pattern Analysis and Machine Intelligence
Pattern Recognition and Machine Learning (Information Science and Statistics)

Pattern Recognition and Machine Learning (Information Science and Statistics)
LIBSVM: A library for support vector machines

ACM Transactions on Intelligent Systems and Technology (TIST)
A real-time automated system for the recognition of human facial expressions

IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper focuses on the variation of EEG at different emotional states. We use pure music segments as stimuli to evoke the exciting or relaxing emotions of subjects. EEG power spectrum is adopted to form features, power spectrum, differential asymmetry, and rational asymmetry. A linear dynamic system approach is applied to smooth the feature sequence. Minimal-redundancy-maximal-relevance algorithm and principal component analysis are used to reduce the dimension of features. We evaluate the performance of support vector machine, k-nearest neighbor classifiers and least-squares classifiers. The accuracy of our proposed method reaches 81.03% on average. And we show that the frequency bands, beta and theta, perform better than other frequency bands in the task of emotion recognition.