Classification of Synchronized Brainwave Recordings using Machine Learning and Deep Learning Approaches

Abstract

It is important to identify and to classify brain signals to diagnose brain diseases. This study uses Synchronized Brainwave Recordings or Electro Encephalography (EEG) signals data available from the University of California, Berkeley, School of Information, to understand features and to classify signals into eight different classes. First, Fast Fourier Transform (FFT) is used for feature extraction and then classifiers like Random Forest, Gradient Boost, Xgboost, Ensemble Voting and Logistic Regression are used to classify the signals. Next, the challenges in classifying using deep learning based approaches like Convolutional Neural Network (CNN) for multi-class classification are discussed.