Özet:

Standing visual stimulated potential (DDGUP) is often used in EEG studies because they have quite high signal-bullying rates and information transfer speed compared to other brain computer interface (BBA) techniques. Standing state paradigms are also used to characterize the preferred frequencies in the dynamic neocortex processes. DDGUPs with a short training period play an important role in practical applications. Signal processing algorithms, used in converting signals into commands, are crucial to improving the performance of BBA systems. In addition, there are very few studies that investigate the classification of DDGUP signals by different methods. In this study, analyses were made on the data set obtained by open access from the internet (AVI SSVEP Dataset). The EEG records in the data set were recorded when the participants looked at a box flaming at seven different frequencies, which quickly changes from black to white. DDGUP signals recorded from the oxypital area are first applied to the Fast Fourier Conversion, divided into the signals subband (delta, teta, alpha, beta and gamma). For each of the subbands, energy and variance properties vectors have been removed. Qualifications are classified by six basic classifiers (LDA, k-NN, SVM, Naive Bayes, Community Learning, Decision Tree). The performance is compared to each other. Classification was analyzed by extracting accuracy values from the 5-watt cross-confirmation model and error matrix. Participants individually considered, up to 100% classification success is achieved in SVM and k-NN classifiers, while the highest success according to the average was achieved in Community Learning classifiers at 79.73.

Anahtar Kelimeler: