基于加权 <i>K</i>-阶传播数的情绪脑网络分类研究

钱宇同; 沈健; 张家祯; 何谈沁; 黄丽亚

doi:10.7507/1001-5515.201905039

基于加权 K-阶传播数的情绪脑网络分类研究

doi: 10.7507/1001-5515.201905039

详细信息

通讯作者:
黄丽亚，Email：huangly@njupt.edu.cn

计量
- 文章访问数: 183
- HTML全文浏览量: 86
- PDF下载量: 0
- 被引次数: 0
出版历程
- 收稿日期: 2019-05-14
- 修回日期: 2019-12-18
- 发布日期: 2020-03-17

Classification of emotional brain networks based on weighted K-order propagation number

More Information

Corresponding author: HUANG Liya, Email: huangly@njupt.edu.cn

摘要

摘要: 脑电信号与人类情绪具有强相关性，情绪脑网络的节点重要性研究为分析情绪脑机制提供了有效手段。本文采用一种新的节点重要性排序方法——加权 K-阶传播数法，设计实现了一种情绪脑网络的分类算法。首先基于 DEAP 情绪脑电数据构建互样本熵脑网络，对正、负情绪下的脑网络分别进行节点重要性排序，以获得多阈值尺度下的特征矩阵。然后通过特征提取和支持向量机实现对情绪的二分类，分类准确率达到 83.6%。结果表明采用加权 K-阶传播数法提取脑网络节点重要性特征进行情绪分类研究是有效的，为复杂网络的特征提取和分析提供了一种新的方法。
- 脑网络 /
- 加权 K-阶传播数 /
- 节点重要性 /
- 特征提取 /
- 支持向量机
Abstract: Electroencephalography (EEG) signals are strongly correlated with human emotions. The importance of nodes in the emotional brain network provides an effective means to analyze the emotional brain mechanism. In this paper, a new ranking method of node importance, weighted K-order propagation number method, was used to design and implement a classification algorithm for emotional brain networks. Firstly, based on DEAP emotional EEG data, a cross-sample entropy brain network was constructed, and the importance of nodes in positive and negative emotional brain networks was sorted to obtain the feature matrix under multi-threshold scales. Secondly, feature extraction and support vector machine (SVM) were used to classify emotion. The classification accuracy was 83.6%. The results show that it is effective to use the weighted K-order propagation number method to extract the importance characteristics of brain network nodes for emotion classification, which provides a new means for feature extraction and analysis of complex networks.
- brain network /
- weighted K-order propagation number /
- importance of nodes /
- feature extraction /
- support vector machine

HTML全文

图 1 加权 K-阶传播数的情绪脑网络分类算法步骤

Figure 1. The flow chart of emotional brain network classification algorithm based on weighted K-order propagation number

下载: 全尺寸图片幻灯片

图 2 矩阵 C 的奇异值分解过程

Figure 2. Singular value decomposition process of matrix C

下载: 全尺寸图片幻灯片

图 3 DEAP 数据集导联位置图

Figure 3. The channel location of the DEAP dataset

下载: 全尺寸图片幻灯片

图 4 负向情绪脑网络

a. 邻接矩阵 A；b. 0-1 矩阵 B

Figure 4. Negative emotional brain network

a. adjacency matrix A; b. 0-1matrix B

下载: 全尺寸图片幻灯片

图 5 某被试正向情绪脑网络在 T₃、T₁₀、T₁₅、T₁₈ 阈值下的网络拓扑图

Figure 5. Network topology of a positive emotional brain network under the threshold T₃, T₁₀, T₁₅, and T₁₈

下载: 全尺寸图片幻灯片

图 6 某被试负向情绪脑网络在 T₃、T₁₀、T₁₅、T₁₈ 阈值下的网络拓扑图

Figure 6. Network topology of a negative emotional brain network under the threshold T₃, T₁₀, T₁₅, and T₁₈

下载: 全尺寸图片幻灯片

图 7 某被试在 T₁₃下的脑网络节点重要性示意图

Figure 7. The importance map of the brain network node corresponding to a subject under the threshold T₁₃

下载: 全尺寸图片幻灯片

表 1 每名被试的 DEAP 数据集的数据组成

Table 1. DEAP dataset representation for each subject

类别	数据维度	数据意义
情绪数据	$40 \times 40 \times 8\;064$	视频 × 导联 × 采样点
标签	$40 \times 4$	视频 × 标签（效价、唤醒度、喜欢度、优势度）

下载: 导出CSV

表 2 不同模型分类准确率结果比较

Table 2. Comparison between classification accuracies of different models

模型	SVD	PCA
KNN	61.8%	66.2%
决策树	51.5%	58.8%
SVM	75.3%	83.6%

下载: 导出CSV

表 3 情绪二分类准确率结果比较

Table 3. Comparison between classification accuracies of our models and previous research for 2 classes

文献	特征	模型	准确率
Candra 等^[2]	小波熵	SVM	65.1%
Tripathi 等^[3]	均值、中位数、最大值等	CNN	81.4%
Tripathi 等^[3]	均值、中位数、最大值等	DNN	75.8%
本文	基于加权 K-阶传播数节点重要性	SVM	83.6%

下载: 导出CSV

参考文献(15)

[1]	Mohammadpour M, Hashemi S M R, Houshmand N. Classification of EEG-based emotion for BCI applications//2017 Artificial Intelligence and Robotics (IRANOPEN). Qazvin: IEEE, 2017: 127-131.
[2]	Candra H, Yuwono M, Chai R, et al. Investigation of window size in classification of EEG-emotion signal with wavelet entropy and support vector machine//The 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. Milan: IEEE, 2015: 1-4.
[3]	Tripathi S, Acharya S, Sharma R D, et al. Using deep and convolutional neural networks for accurate emotion classification on DEAP dataset//Proceedings of the Twenty-Ninth AAAI Conference on Innovative Applications. Palo Alto: AAAI Press, 2017: 4746-4752.
[4]	Wang N, Wang Y, Li Y, et al. Gamma oscillation in brain connectivity in emotion recognition by Granger causality//2011 4th International Conference on Biomedical Engineering and Informatics (BMEI). Shanghai: IEEE, 2011: 762-766.
[5]	Richman J S, Moorman J R. Physiological time-series analysis using approximate entropy and sample entropy. Am J Physiol Heart Circ Physiol, 2000, 278(6): H2039-H2049. doi: 10.1152/ajpheart.2000.278.6.H2039
[6]	Pincus S M, Viscarello R R. Approximate entropy: a regularity measure for fetal heart rate analysis. Obstet Gynecol, 1992, 79(2): 249-255.
[7]	Ledoux J E. Emotion circuits in the brain. Annu Rev Neurosci, 2009, 23(23): 155-184.
[8]	汪小帆, 李翔, 陈关荣. 复杂网络理论及其应用. 北京: 清华大学出版社, 2006.
[9]	Kitsak M, Gallos L K, Havlin S, et al. Identification of influential spreaders in complex networks. Nat Phys, 2010, 6(11): 888-893. doi: 10.1038/nphys1746
[10]	Berkhin P. A survey on pagerank computing. Internet Math, 2005, 2(1): 73-120. doi: 10.1080/15427951.2005.10129098
[11]	黄丽亚, 霍宥良, 王青, 等. 基于K-阶结构熵的网络异构性研究. 物理学报, 2019, 68(1): 325-336.
[12]	黄丽亚, 汤平川, 霍宥良, 等. 基于加权K-阶传播数的节点重要性研究. 物理学报, 2019, 68(12): 311-321.
[13]	Diego Rodriguez J, Perez A, Antonio Lozano J. Sensitivity analysis of k-fold cross validation in prediction error estimation. IEEE Trans Pattern Anal Mach Intell, 2010, 32(3): 569-575. doi: 10.1109/TPAMI.2009.187
[14]	Koelstra S, Muhl C, Soleymani M, et al. DEAP: A database for emotion analysis using physiological signals. IEEE Trans Affect Comput, 2012, 3(1): 18-31. doi: 10.1109/T-AFFC.2011.15
[15]	Russell J A, Lewicka M, Niit T. A cross-cultural study of a circumplex model of affect. J Pers Soc Psychol, 1989, 57(5): 848-856. doi: 10.1037/0022-3514.57.5.848