Performance Analysis of Machine Learning Algorithms for Classifying Hand Motion-Based EEG Brain Signals

Brain-computer interfaces (BCIs) records brain activity using electroencephalogram (EEG) headsets in the form of EEG signals; these signals can be recorded, processed and classified into different hand movements, which used to control other IoT devices. Classification movements will one step closer applying algorithms real-life situations headsets. This paper uses feature extraction techniques sophisticated machine learning classify from prosthetic hands for amputated persons. To achieve good classification accuracy, denoising is a significant step. We saw considerable increase all models when moving average filter was applied raw data. Feature like fast fourier transform (FFT) continuous wave (CWT) were this study; three types features extracted, i.e., FFT Features, CWT Coefficients scalogram images. trained compared (ML) logistic regression, random forest, k-nearest neighbors (KNN), light gradient boosting (GBM) XG boost on deep (DL) VGG-16, DenseNet201 ResNet50 Boost with gave maximum accuracy 88%.