Information Detection in Brain Using Wavelet Features and K-Nearest Neighbor

Brain signals provide information about the various activities happening in brain at the time of occurrence of different events. Information specifically known to a person can be assessed using electroencephalogram (EEG) signals. Recognition of specific domain knowledge can be useful for screening individuals by job placement and law enforcement agencies. An individual possessing specific domain knowledge in real life can benefit the society with his/her knowledge and can be distinguished from the individuals possessing no domain knowledge. There is a huge demand of computer science professionals in information technology organizations, government institutions, security agencies, research laboratories, and private organizations. Hence, in the present work it is proposed to detect the knowledge of computer science in the subjects using brain signals. The text stimuli related to computer science domain was presented to the participants on a computer screen. Wavelet decomposition was used to extract features in the delta frequency band of EEG signals. Principal component analysis (PCA) gives reduced dimensionality of wavelet features and leads to better classification than the original features. It is observed that k-nearest neighbor (k-NN) classifier performs significantly better than other classifiers with a classification accuracy of 80%. These results show the efficacy and easy implementation of wavelet features with k-NN classifier in effectively recognizing the experts of a particular field.