Emotion Recognition of EEG Signals Based on the Ensemble Learning Method: AdaBoost

In recent years, with the continuous development of artificial intelligence and brain-computer interface technology, emotion recognition based on physiological signals, especially, electroencephalogram (EEG) has become a popular research topic attracted wide attention. However, how to extract effective features from EEG signals accurately recognize them by classifiers have also an increasingly important task. Therefore, in this paper, we propose method ensemble learning method, AdaBoost. First, consider time domain, time-frequency nonlinear related emotion, preprocessed fuse into eigenvector matrix. Then, linear discriminant analysis feature selection is used reduce dimensionality features. Next, use optimized sets train classifier AdaBoost, for binary classification. Finally, proposed been tested DEAP data set four emotional dimensions: valence, arousal, dominance, liking. The proved be recognition, best average accuracy rate can reach up 88.70% dominance dimension. Compared other existing methods, performance significantly improved.