基于时频域组合特征的脑电信号情感分类算法

为了提高基于脑电信号(electroencephalogram, EEG)情感识别的准确率,提取了脑电信号的时域与频域特征,并且将其进行组合形成时频域组合特征,作为不同识别模型下的输入。采用集成决策树(bagging tree, BT)、贝叶斯线性分析(Bayesian linear discriminant analysis, BLDA)、线性判别分析(linear discriminant analysis, LDA)及支持向量机(support vector machine, SVM)四种浅层机器学习算法对EEG在效价与唤醒度上进行二分类情感识别。实验结果表明,DEAP数据集在效价上,基于时频域组合特征在BT分类器下的识别精度平均达到92.54%,在唤醒度维度上基于时频域组合特征在SVM下平均识别精度达到94.62%。

EEG Emotion Classification Algorithm Based on Combined Features in Time and Frequency Domain

In order to improve the accuracy of emotion recognition based on Electroencephalogram (EEG), and to compare the effect of emotion recognition of EEG signals under different recognition models. By Bagging tree(BT) , bayesian linear analysis, linear discriminant analysis and SVM four types of shallow machine learning algorithm for EEG in valence and arousal binary classification emotion recognition .Experiment results show that in DEAP data set, each model on the combination of time and frequency domain characteristics of average recognition auc is higher, which based on the classification effect under the BT classifier, average valence and arousal dimension auc reached 92.54% and 92.54% respectively.