The effects of EEG feature extraction using multi-wavelet decomposition for mental tasks classification

In modern life, the identification system is considered as one of the most challenging projects because identity authentication needs to be secure. Researchers have developed digital authentication techniques which are implemented in society. One of these techniques is using biometric technology which is commonly known as face recognition, voice recognition, and fingerprinting. These techniques have achieved a high level of authentication but are subject to hacking or counterfeiting. In this paper, a new identification method based on electroencephalogram (EEG) signals is proposed. The EEG method uses a standard EEG database which deals with five different thought patterns or mental tasks which are multiplication, baseline, letter composition, rotation, and visual board counting. Using ANN (artificial neural network) classifier, EEG signals were classified. The performance of this proposed method is evaluated using five criteria: (accuracy, sensitivity, specificity, F-Score measure, and false acceptance rate). The experimental results show that the EEG features extraction with wavelet 10 decomposition levels can achieve better than 5 decomposition levels for all mental tasks. The proposed method achieved the highest accuracy when using a visual counting mental task.