A Convolutional Neural Network Feature Fusion Framework with Ensemble Learning for EEG-based Emotion Classification

In recent years, mental health has received more and more attention. With the development of artificial intelligence, machine learning has also been widely used in the field of mental health, e.g., emotion analysis. We propose a feature fusion framework based on convolutional neural network with correlation coefficient matrix and synchronization likelihood matrix of EEG signals for emotion analysis. To further improve the performance, we take the proposed fusion framework as a feature extractor, i.e., taking the output of the layer before softmax as feature, and use stacking strategy for ensemble learning. Experiments on the DEAP database show the effectiveness of the proposed method.