Real-Time On-Chip Machine-Learning-Based Wearable Behind-The-Ear Electroencephalogram Device for Emotion Recognition

In this study, we propose an end-to-end emotion recognition system using ear-electroencephalogram (EEG)-based on-chip device that is enabled the machine-learning model. The has integrated gathers EEG signals from electrodes positioned behind ear; it more practical than conventional scalp-EEG method. relative power spectral density (PSD), which feature used in derived fast Fourier transform over five frequency bands. Directly on embedded device, data preprocessing and extraction were carried out. Three standard machine learning models, namely, support vector (SVM), multilayer perceptron (MLP), one-dimensional convolutional neural network (1D-CNN), trained these rich classification features. traditional approach, integrates a model into application software personal computer (PC), cumbersome lacks mobility, makes challenging to use real-life applications. Besides, PC-based not sufficiently real-time because of connection latency acquisition device. To overcome limitations, wearable capable performing signal processing immediately after task for result. order perform prediction emotions, 1D-CNN was chosen as pre-trained PSD characteristics input based evaluation set results. Additionally, developed smartphone alerted user whenever negative emotional state identified displayed information real life. Our test results demonstrated feasibility practicability our recognition.