Study On Recognition Of Mental Tasks From Electroencephalogram Recordings

The "Brain-Computer Interface (BCI)" system is aimed at developing a technical system with extra communication and control channels that do not depend on the brain's normal output pathways of peripheral nerves and muscles. During these years, the BCI projects that based on the electroencephalogram (EEG) signals have been the key topics. This paper focus on some topics described as following based on the summarizer of:(1) Feature extraction of the EEGAnalyzing the origin, the mechanism and the framework of the EEG and using the power spectral density, AR model parameter, band power and the wavelet entropy as the features of the EEG recorded during imagination of hand movements. Moreover, we compared the performance of these features.(2) Application of the phase synchronization in the EEGSensory stimulation, motor behavior, and mental imagery can change the functional connectivity within the cortex and results in an amplitude suppression "event-related desynchronization (ERD)" or in an amplitude enhancement "event-relatedsynchronization(ERS)" of μ and βrhythms. Based on this characteristic, we applythe concept of phase synchronization and the statistical distribution of the relative instantaneous phase to the EEG as the signals feature to classify the movement consciousness. With the experiments we have approved the feasibility and the validity of this phase synchronization information in the BCI system.(3) Linear discriminant analysis (LDA) algorithmLinear classification methods, like Linear Discriminant Analysis (LDA), can obtain a reliable classifier by fewer samples. However, sometimes the LDA output seemed to be biased towards one class. Here we used LDA algorithm in classify of the EEG and improved it. We used the Mahalanobis distance based classifier (MDBC) method to distinguish the left and right hand motor imaginary tasks and gained the preferably result.(4) Support vector machineRecent advances in machine learning research have pointed out the advantages of support vector machines (SVM) over other classification techniques. Solid theoretical foundations, good generalization capabilities and easy parameters updating are among the most appealing qualities of SVM for BCI applications. Here we applied the SVM theory to the classification of the EEG and gained the result similar to the LDA.