Implementing eigen features methods/neural network for EEG signal analysis
Date
2013-01Author
Saidatul Ardeenawatie, Awang
Pandiyan, Paulraj Murugesa, Prof. Dr.
Sazali, Yaacob, Prof. Dr.
Metadata
Show full item recordAbstract
This paper presented the possibility of implementing eigenvector methods to represent the features of electroencephalogram signal. In this study, three eigenvector methods were investigated namely Pisarenko, Multiple Signal Classification (MUSIC) and Modified Covariance. The ability of the features in representing good character of signal in order to discriminate two different EEG signals for relaxation and writing signal were tested using neural network. The power level obtained by eigenvector methods of the EEG signals were used as inputs of the neural network trained with Levenberg-Marquardt algorithm. The classification result shows that Modified Covariance method is a better technique to extract features for relaxation-writing task.
URI
http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6481149http://dspace.unimap.edu.my:80/dspace/handle/123456789/34157
Collections
- Conference Papers [2600]
- Sazali Yaacob, Prof. Dr. [250]
- Paulraj Murugesa Pandiyan, Assoc. Prof. Dr. [113]