An Extensive Review of Machine Learning Techniques for EEG Signal Processing

Electrical brain activity is detected by signals in an Electroencephalogram (EEG). Based on their frequencies, EEG signals are usually put into one of five groups: delta, theta, alpha, beta, and gamma. These signals help find a pattern that can be used to predict when a person will have a seizure. Classifying a seizure is a very important job for a doctor, as it helps them figure out what kind of seizure it is and if there will be any other problems. The goal of seizure classification is to learn as much as possible about the EEG signals. Literature shows that there are a lot of EEG signal pre-processing techniques, selection methods, feature extraction from EEG signals, and classification algorithms that can be used to find out if someone is having a seizure. The most important thing about pre-processing EEG signals is that it improves the quality of raw experimental data, which leads to better datasets, better classifications, and better accuracy. This study gives an overview of recent EEG pre-processing methods, datasets that can be used for experiments, and EEG classification techniques that will help a beginner researcher build on and use the right techniques.