A neuro-fuzzy based approach for resting-state detection using a consumer-grade EEG

This work aims to introduce a methodology for resting-state brain activities detection by a consumer-grade EEG. From one hand, an adaptive noise reduction methodology based on non-linear Principal Component Analysis Neural Network is adopted. On the other hand, a Neuro-Fuzzy model (i.e., Fuzzy Relational Neural Network) is considered for brain activities detection since a combination of neural networks and fuzzy technology enhances the performance of control, decision-making and data analysis systems. Experiments are made on a corpus containing the activation strength of the fourteen electrodes of an EEG headset for eye state detection. We proved that by using the noised signals, the proposed methodology permits to obtain a high rate of classification accuracy.