A New Method for EEG Signals Classification Based on RBF NN

EasyChair Preprint no. 8786

Abstract

Traditional EEG reviews are tedious and time-consuming, particularly the outpatient form, so automation is required. The researchers concentrated on the designing of a three-class EEG classifier utilizing extracting the features (FeExt) and Radial Basis Functional Neural Network (RBFnn) for this manuscript. To identify the trends equally, RBFnn can be trained if FeExt is completed. Depending on the EEG signal, many types of anomalies can be detected, and a seizure signal is one of them. The three types of EEG signals are stable, interactive, and seizure signals. The aim of this manuscript is to classify EEG signals RBFnn-based. CHB-MIT Scalp EEG dataset was relied on for EEG signal data taken. There are 55- various FeExt schemes are tested and a relatively accurate and rapid classifier is built. The extraction methods were not discussed or compared to the literature with 10 morphological properties. Based on findings, thebest classifier topology is considered to be the multilayer perceptron with momentum learning law, while the stronger performance of FeExt techniques is: PCA, Bi-gonal 2.2, coif1, DCT, db9, Re-Bi-gonal 1.1, and sym2. The documented outcomes can be utilized efficiently classification for EEG rhythm for analysis rapidly by a Neurologist specialist. Therefore, time-saving and fast and careful diagnosis. The EEG rhythm classification for more cerebral diseases can be utilized with a similar procedure

Keyphrases: EEG, EEG rhythm Classifier, RBFNN

@Booklet{EasyChair:8786,
  author = {Shokhan M. Al-Barzinji and Mohanad A. Al-Askari and Azmi Shawkat Abdulbaqi},
  title = {A New Method for EEG Signals Classification Based on RBF NN},
  howpublished = {EasyChair Preprint no. 8786},

  year = {EasyChair, 2022}}