Static and dynamic activities prediction of human using machine and deep learning models

Abstract

Every behavior displays a specific signature encoded in an electrical signal produced inside the brain by mV. An electroencephalogram (EEG) is a test used to detect electrical activity in your brain using small metal disks (electrodes) attached to your scalp. In this research paper, FFT-based mental arithmetic task-based feature extraction, EEG data collection is introduced and extracted features are provided as input to classify EEG signals by the Convolutional Neural Network (CNN) classifier. To achieve this, first the automated filters are used to noise the signal, and then feature vectors are extracted using a technique of feature extraction. There are many techniques for extracting features, such as Fast Fourier Transform (FFT), Eigenvector Methods (EM), different types of Wavelet Transform like Discrete Wavelet Transform, Continuous Wavelet transform, Time-Frequency Distribution (TFD), Autoregressive Method (ARM), etc. Fast Fourier transform is used in this project. After applying FFT over signals that have feature vectors as outputs, the feature vector is run on the classifier to construct the model to determine that the output signal will be while or before the mental arithmetic function is performed. FFT (fast Fourier transform) is used to extract features and CNN to classify EEG signals. The proposed model has achieved an accuracy of 96.81%, which demonstrates the superiority of our model.