Abstract

Electroencephalogram (EEG) is a signal that provides information about the brain and nerve function. EEG signals have low amplitude and do not have a specific pattern, so it is not easy to be analyzed. Based on existing research, the frequency of the human brain is different for each phase of conscious, relaxed, light sleep, deep sleep, panic, and so on. Several studies have been conducted for the classification of EEG signals. This research build a classification system based of EEG signals using the coefficient of autoregressive (AR) and Artificial Neural Network (ANN) with back propagation learning method. Classification in this research were divided into four, namely to distinguish the signal pattern when someone want to hold, want to release, was holding, and was taking off. The result show the use of AR coefficient can improve the average classification accuracy. With the EEG signal classification systems using AR coefficient and ANN back propagation obtained an average accuracy of 60% for data training and 57.5% for data test with all condition. Classified system has been applied in the software so easy to public used.

Keywords : EEG sigmals, AR coefficient, back propagation