Mind-Controlled Web Browser Navigation Based on Brain Computer Interfaces

Brain-Computer Interfaces (BCIs) are communication systems that enable direct interaction between the human brain and machines or devices without the need for physical contact, utilizing EEG signals generated from brain activity. A BCI system typically involves two key stages: feature extraction and classification. The classification process relies on signals collected from specific EEG sensor groups. One of the main challenges in classifying motor imagery EEG signals arises from the fact that EEG data is often a mixture of meaningful signals and noise. As a result, selecting an effective classification technique is crucial in EEG-based BCIs. In this study, we applied machine learning and deep learning classifiers to categorize motor imagery in EEG data used for BCIs. We assessed the performance of several classification techniques, including Support Vector Machine (SVM), Multilayer Perceptron Neural Network (MLP-ANN), Decision Tree (DT), and Random Forest (RF), as these are commonly employed in classification tasks. The results indicated that Random Forest (RF) outperformed the other methods, achieving the highest accuracy of 82.72%. This data was subsequently used to develop a web browser navigable via brain signals.