EEG Signal with Feature Extraction using SVM and ICA Classifiers

1. INTRODUCTIONgrams (EEGs) are recordings of the electricalpotentials developed by the brain. Analysis of EEG activity hasbeen achieved principally in clinical settings to identify pathologiesand epilepsies. An interpretationof the EEG is used to visual inspection by a neurophysiologist. EEG technology used many electrodes on thehuman skull, such signals givesinformation indirectly about physiological functions, which are related tothe brain, these signals are verynumerous. The EEG integrated technical devices with embedded intelligence andit allows for Brain-Computer- Interfaces (BCI) to analysis EEG design. BCI is composed of signalcollection and processing, pattern identification and control systems.EEG classification has many number of features, it comes from the fact that are, (i) EEG signals are non-stationary, thus, features must be computedin a time- varying manner, and (ii) Number of EEG channelsis large. For the classification process, a multilayer perceptron (MLP) neural network is trained with the back propagationalgorithm.