Statistical Comparison of Classification Methods in EEG Signals


EKİM G. , ATASOY A. , İKİZLER N.

25th Signal Processing and Communications Applications Conference (SIU), Antalya, Turkey, 15 - 18 May 2017 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Volume:
  • Doi Number: 10.1109/siu.2017.7960182
  • City: Antalya
  • Country: Turkey

Abstract

EEG is a test method that contains important information about brain activity and it is frequently used in the diagnosis and treatment of brain diseases. In this study, the EEG dataset from the University of Bonn, Department of Epileptology Database was used. First, spectral analysis of EEG records was performed with discrete wavelet transform. Then, these records were classified using Naive Bayes, K-Nearest Neighbor, Support Vector Machines and Decision Trees methods, and statistically compared with the results that has been found. In term of classification success and algorithm speed, it has been determined that the best method is the K-Nearest Neighbor method.