EEG-based recognition of attention state using wavelet and support vector machine

Recognition of attention level is required to the evaluation of student learning process and employee development. This research proposed a recognition of attention state using wavelet filter and Support Vector Machine (SVM). It was divided by two states, that attention and inattention. EEG signal was extracted using Wavelet that filtered in the frequency range of 5-30 Hz. It is relevant with information of EEG signal against attention, that theta, alpha and beta waves. Therefore, using wavelet filter increased accuracy from 44-58% into 69-83%. Using SVM is simpler than using Neural Networks therefore could lead to convergence and shorter training time. Therefore, the system can be developed with a large training data. The classification system trained with four subjects to obtain the separation coefficient of linear hyperplane of SVM. It took less than one minute. Testing of four subjects gave an accuracy of 77-83%. The system that has been implemented into a software.