A Comparative Study of Classification Techniques for P300 Speller
VaishaliPatelia1, Maulika S. Patel2
1VaishaliPatelia, Department of Computer Engineering, G. H. Patel College of Engineering & Technology, Gujarat Technological University, V.V. Nagar, (Gujarat), India.
2Maulika S. Patel, Professor & Head, Department of Computer Engineering, G. H. Patel College of Engineering and Technology, Gujarat Technological University, V.V. Nagar, (Gujarat), India.
Manuscript received on 27 April 2020 | Revised Manuscript received on 09 May 2020 | Manuscript Published on 22 May 2020 | PP: 102-106 | Volume-9 Issue-7S July 2020 | Retrieval Number: 100.1/ijitee.G10200597S20 | DOI: 10.35940/ijitee.G1020.0597S20
Open Access | Editorial and Publishing Policies | Cite | Zenodo | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: P300 speller in Brain Computer Interface (BCI) allows locked-in or completely paralyzed patients to communicate with humans. To achieve the performance of characterization and increase accuracy, machine learning techniques are used. The study is about an event related potential (ERP) P300 signal detection and classification using various machine learning algorithms. Linear Discriminant Analysis (LDA) and Support Vector Machine (SVM) are used to classify P300 and Non-P300 signal from Electroencephalography (EEG) signal. The performance of the system is evaluated based on f1-score using BCI competition III dataset II. In our system, we used LDA and SVM classification algorithms. Both the classifiers gave 91.0% classification accuracy.
Keywords: Brain Computer Interface, P300 Speller, Event Related Potential, Linear Discriminant Analysis, Support Vector Machine.
Scope of the Article: Classification