Iranian Journal of Biomedical Engineering (IJBME)
نوع مقاله : مقاله کامل پژوهشی
نویسندگان
1 عضو هیات علمی
2 گروه بیوتکنولوژی، دانشکده علوم و فناوریهای نوین، دانشگاه سمنان، سمنان، ایران
کلیدواژهها
عنوان مقاله English
نویسندگان English
Biometric authentication based on brain signals, especially in steady-state visually evoked potential (SSVEP)- based approaches, is regarded as a robust, spoof-resistant, and reliable method in security systems and brain–computer interface applications. In this study, an authentication process for SSVEP signals was performed using spatial feature extraction and classification. To this end, using the Benchmark dataset including 35 subjects, a feature-extraction framework based on multiset canonical correlation analysis (Mset-CCA) was designed and implemented to effectively extract stable and discriminative patterns of brain activity for personal identity recognition. Classification was performed using linear discriminant analysis (LDA), resulting in an accuracy of 99.43% in identity recognition. For comparison, the results obtained using several other classifiers, including support vector machine, k-nearest neighbor, ensemble learning, decision tree, and naive Bayes, were also computed and reported. The findings demonstrated that the proposed method, based on Mset-CCA and the LDA classifier, substantially improved the authentication system’s performance compared to state-of-the-art methods
کلیدواژهها English
