Word prediction from fMRI data based on C-SVC and a series classifier
7th International Conference on Computer and Knowledge Engineering, ICCKE 2017, Mashhad, İran, 26 - 27 Ekim 2017, cilt.2017-January, ss.131-136, (Tam Metin Bildiri)
- Yayın Türü: Bildiri / Tam Metin Bildiri
- Cilt numarası: 2017-January
- Doi Numarası: 10.1109/iccke.2017.8167941
- Basıldığı Şehir: Mashhad
- Basıldığı Ülke: İran
- Sayfa Sayıları: ss.131-136
- Anahtar Kelimeler: AdaBoost M2, classification, C-SVC, fMRI data, KNN, PCA, voxel, word prediction
- Atatürk Üniversitesi Adresli: Hayır
Özet
Word prediction is an applicable task for medical purposes and it can be done by analyzing brain's activities. Functional Magnetic Resonance Imaging (fMRI) is a technique for obtaining 3D images, related to the neural activity of brain through time. By subtracting fMRI images, which are captured consecutively, brain's operation can be detected. In this paper, a novel approach, based on machine learning algorithms, is designed to predict words from fMRI data. In the proposed method, after dimensionality reduction by means of principal component analysis (PCA), C-support vector classification (C-SVC) and a series classifier are applied for fMRI data classification. Results of the proposed method is compared with other classification approaches. Experiments show that the proposed method increases precisian of fMRI data classification and word prediction reliably and incredibly.