A semi-supervised machine learning algorithm developed by the combination of the concepts of electrical brain signal power analysis, modified k-means, and higher-order statistics.
Extensive analysis of time series brain-signal data leading to the development of a data-driven machine learning algorithm with the help of signal power calculation and utilization of randomized statistical learning.
Use of brain microstate analysis of the brain signal data i.e. Electroencephalographic (EEG) data.
Deployment of the unsupervised machine learning algorithm k-means to cluster the unlabelled EEG data.
Building up a modular-based code repository pipeline for exploratory signal-power analysis of the EEG datasets using libraries like MNE-Python, Python, NumPy, Sci-py.
Detection of muscle artifacts (noise) in the EEG data using signal-power analysis.
Removal of muscle artifacts using the technique of fit-back of brain microstates and higher statistical learning e.g. randomization statistics.
Evaluation of the algorithm with the famous Multiple Artifact Rejection Algorithm (MARA) plugin in MATLAB.