Özet:

Electroencephalography (EEG) is a method that allows measurement of brain activity and is non-initiative. According to the electrophysiological source, there are seven different types of EEG recordings available. Among them are the sights that are accompanied. Recently, the stable state of visual alarming potential (DDGUP) has often begun to be used in brain computer interfaces (BBAs). In literature, the DDGUP signals are focused on the detection of the objects the examiners focus on. However, not only the focused object information, but also the frequency of the alert is entered into the DDGUP signal. Additionally, the DDGUP signals are not stable because the wave conversion is an effective method for the output of authenticity. However, it is necessary to determine the main wave type to be used before the wave conversion. In this study, the basic properties extracted from different main wave types for the determination of the alert frequency from the DDGUP signal were applied to different classifiers. From the well-known DDGUP signals recorded at eight different alarm frequencies, properties such as energy, variance and entropy in the EEG frequency bands were extracted. Six different wave functions (Haar, Daubechies, Symlet, Coiflet, Biorthogonal, Reverse Biorthogonal) have been used for this purpose. Also, a variance analysis (ANOVA) has been conducted to identify those that are effective from a large number of properties. The DDGUP signals are classified with six basic classifiers and the classification performance is compared with each other. According to the results obtained, ANOVA has shown that the preference selection has reduced its classification performance. Furthermore, it has been seen that the highest achievement for each applicant is achieved in different classifiers and main wider types when taken into consideration. When all the indicators are evaluated together, the algorithm of community classification (Ensemble learner) and the properties obtained using the Reverse Biorthogonal main wider are used together, the alert frequency can be detected with the highest accuracy. As a result, it is believed that BBA devices that work with higher accuracy could be designed if the alert frequency information is extracted from the SSVEP.

Anahtar Kelimeler: