EEG feature extraction for classifying emotions using FCM and FKM

  • Authors:

  • M. Murugappan;M. Rizon;R. Nagarajan;S. Yaacob

  • Affiliations:

  • School of Mechatronics Engineering, Universiti Malaysia Perlis, Kangar, Perlis, Malaysia;School of Mechatronics Engineering, Universiti Malaysia Perlis, Kangar, Perlis, Malaysia;School of Mechatronics Engineering, Universiti Malaysia Perlis, Kangar, Perlis, Malaysia;School of Mechatronics Engineering, Universiti Malaysia Perlis, Kangar, Perlis, Malaysia

  • Venue:
  • ACACOS'08 Proceedings of the 7th WSEAS International Conference on Applied Computer and Applied Computational Science
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

The Electroencephalogram (EEG) is one of the useful biosignals detect the human emotions. This paper discusses on a research conducted to determine the changes in the electrical activity of the human brain related to distinct emotions. We designed a competent acquisition protocol for acquiring the EEG signals under audio-visual induction environment. The EEG data has been collected from 6 healthy subjects with in an age group of 21-27 using 63 biosensors. From the subjective analysis on each emotion, three emotions have been identified with higher agreement. After preprocessing the signals, discrete wavelet transform is employed to extract the EEG parameters. The feature vectors derived from the above feature extraction method on 63 biosensors form an input matrix for emotion classification. In this work, we have used Fuzzy C-Means (FCM) and Fuzzy k-Means (FKM) clustering methods for classifying the emotions. We have also analyzed the performance of FCM and FKM on reduced number of 24 biosensors model. Finally, we compared the performance of clustering the discrete emotions using FCM and FKM on both 64 biosensors and 24 biosensors. Results confirm the possibility of using wavelet transform based feature extraction for assessing the human emotions from EEG signal, and of selecting a minimal number of channels for emotion recognition experiment.