Machine Learning Methods for Electroencephalogram (EEG) Big Data in the Context of MIOT Smart Systems

Abstract

This book chapter presents several aspects concerning the deep learning (DL) techniques for big data obtained by EEG (electroencephalogram) investigation using smart systems in the context of the medical Internet of Things (MIOT) environment. The learning machine algorithms used in the classification of the signals specific to chronic diseases such as epilepsy, neurological disorders (neuropathy, myopathy), brain-computer interfacing, sleep, and cognitive monitoring are beneficial for EEG classification. DL, as a machine learning method, allows the users to obtain predicted output based on the input data that are subject to training, testing, and validation. Also, at present, this method, extremely used for natural language processing (NLP), computer vision, and speech processing, and for MI-BCI (motor imagery—brain-computer interface) classification, is desirable to be used in the BCI systems for wearable devices for restoring the disabled functions. Moreover, the convolutional neural networks (CNNs), classes of deep neural networks (DNN), are used for decoding the EEG brain signals. The intracranial electroencephalography (iEEG) data used for seizure investigation and prediction of epileptic seizure require DL algorithms to classify and to analyze the signals.