| Summary: Electroencephalogram (EEG) measurements are used in the field of medical research such as determination of the level of anesthesia, treatment of epilepsy, and treatment of diseases associated with apnea and human computer interaction systems.
In this study, a simple Angle-Amplitude transformation method is proposed for classification of EEG signals. Angle-Amplitude transformation is a kind of finite amplitude frequency transformation based on the changing point of a signal (from negative to positive and vice versa). By employing the transformation, arbitrary time domain EEG signals are converted to two dimensional finite images. The attributes of resulting images are exploited in the classification of the EEG signals. Principal Component Analysis (PCA) approach is employed for future extraction on the image; k-Nearest Neighbor (k-NN), Support Vector Machine (SVM), and Artificial Neural Network (ANN) methods are applied for the classification.
The obtained results also show that the EEG signals are quite successfully classified by employing the proposed method with the transformation.
Key Words: Elektroansefalogram, Human Computer Interaction, Principal Component Analysis, Support Vector Machines, K Nearest Neighbour, Transformation, Changing Points, Feature Extraction, Classification |
