摘 要：张量分解作为一种高维数据分析工具能够结合多个模态的信息从而获取具有判别信息的特征，但是在高维空间上进行张量分解存在计算复杂度高的问题。为了解决该问题，研究借助随机奇异值分解速度快的特点，提出基于随机奇异值分解的张量Tucker 分解（张量 R-Tucker 分解），并将其用于 BCICIV2b 数据集的特征提取和分类中。实验结果显示：相比张量 Tucker 分解，张量 R-Tucker 分解特征提取速度提升 22%，并且平均分类准确率达到 80.93%，与现有基于矩阵的方法相比提高 10.12%。
关键词：张量分解；随机奇异值分解；运动想象；脑机接口
DOI:10.19850/j.cnki.2096-4706.2023.02.001
中图分类号：TP301.6 文献标识码：A 文章编号：2096-4706（2023）02-0001-07
Research on BCI Data Classification Based on Tensor R-Tucker Decomposition
ZHANG Shuai
(Taiyuan Normal University, Jinzhong 030600, China)
Abstract: As a high-dimensional data analysis tool, Tensor Decomposition can combine the information of multiple modalities to obtain features with discriminative information, but there is a problem of high computational complexity problem for Tensor decomposition on high dimensional space. In order to solve the problem, this research takes advantage of the b fast speed of Randomized Singular Value Decomposition, and proposes the Tensor Tucker Decomposition (Tensor R-Tucker Decomposition) based on Randomized Singular Value Decomposition, and uses it for feature extraction and classification of the BCICIV2b dataset. The experimental results show that compared with the Tensor-Tucker Decomposition, the feature extraction speed of the Tensor R-Tucker Decomposition is increased by 22%, and the average classification accuracy rate reaches 80.93%, which is 10.12% higher than the existing matrix-based methods.
Keywords: Tensor Decomposition;Randomized Singular Value Decomposition;motor imagery; Brain Computer Interface
参考文献：
[1] LOTTE F，BOUGRAIN L，CICHOCKI A，et al. A Review of Classification Algorithms for EEG-Based Brain–Computer Interfaces：A 10 Year Update [J].Journal of Neural Engineering， 2018，15（3）：031005.
[2] NORCIA A M，APPELBAUM L G，ALES J M，et al. The Steady-State Visual Evoked Potential in Vision Research：A Review [J]. Journal of Vision，2015，15（6）：4.
[3] LUCK S J.An Introduction to the Event-Related Potential Technique [M].Bradford：A Bradford Book，2014.
[4] PFURTSCHELLER G，NEUPER C. Motor Imagery and Direct Brain-Computer Communication [J].Proceedings of the IEEE， 2001，89（7）：1123-1134.
[5] P F U RT S C H E L L E R G . E v e n t - R e l a t e d E E G / M E G Synchronization and Desynchronization：Basic Principles [J]. Neurophysiol Clin，1999，110（11）：1842-1857.
[6] MA Y，ZHENG L S，YI Z K，et al. Short-Time Fourier Transform Covariance and Selection，A Feature Extraction Method for Binary Motor Imagery Classification [C]//2021 IEEE International Conference on Real-Time Computing and Robotics （RCAR）. Xining：IEEE，2021：1316-1322.
[7] YOU Y，CHEN W Z，ZHANG T. Motor Imagery EEG Classification Based on Flexible Analytic Wavelet Transform [J]. Biomedical Signal Processing and Control，2020，62（615）：102069.
[8] PHAN A H，CICHOCKI A. Tensor Decompositions for Feature Extraction and Classification of High Dimensional Datasets [J].Nonlinear Theory and Its Applications，IEICE，2010，1（1）：37-68.
[9] CONG F Y，LIN Q H，KUANG L D，et al. Tensor Decomposition of EEG Signals：A Brief Review [J].Journal of Neuroscience Methods，2015，248：59-69.
[10] ZHANG Q Z，GUO B，KONG W Z，et al. Tensor-Based Dynamic Brain Functional Network for Motor Imagery Classification [J]. Biomed Signal Process Control，2021，69.
[11] ROÁKOVÁ Z，ROSIPAL R，SEIFPOUR S. Tucker Tensor Decomposition of Multi-session EEG Data [C]//29th International Conference on Artificial Neural Networks.Bratislava：Springer， 2020：115-126.
[12] KRISHNAN K K，SOMAN K P. Canonical Polyadic Decomposition of EEG Image Tensor for BCI Applications [M]//ICT Systems and Sustainability.Singapore：Springer，2022：819-826.
[13] 李洁 . 多模态脑电信号分析及脑机接口应用 [D]. 上海：上海交通大学，2009.
[14] BRUNTON S L，KUTZ J N. Data-Driven Science and Engineering：Machine Learning，Dynamical Systems，and Control [M]. Cambridge：Cambridge University Press，2019.
[15] HALKO N，MARTINSSON P-G，SHKOLNISKY Y，et al.An Algorithm for the Principal Component Analysis of Large Data Sets [J].SIAM Journal on Scientific Computing，2011，33（5）： 2580-2594.
[16] ZHANG J N，SAIBABA A K，KILMER M E，et al. A Randomized Tensor Singular Value Decomposition Based on the T-Product [J/OL].arXiv：1609.07086 [math.NA].[2022-08-09].https:// arxiv.org/abs/1609.07086.
[17] KOLDA T G，BADER B W. Tensor Decompositions and Applications [J].SIAM Review，2009，51（3）：455-500.
[18] DE L L，VANDEWALLE J，DE M B. A Multilinear Singular Value Decomposition [J].SIAM Journal on Matrix Analysis and Applications，2000，21（4）：1253-1278.
[19] ANDERSSON C A，BRO R. Improving the Speed of MultiWay Algorithms：Part I.Tucker3 [J].Chemometrics and Intelligent Laboratory Systems，1998，42（1-2）：93-103.
[20] LATHAUWER L D，MOOR B D，VANDEWALLE J. On the Best Rank-1 and Rank-（R1，R2，...，Rn） Approximation of Higher-Order Tensors [J].SIAM Journal on Matrix Analysis and Applications，2006，21（4）：1324-1342.
[21] LEEB R，BRUNNER C，MULLER-PUTZ G R，et al.BCI Competition 2008 Graz data set B[EBOL].[2022-08-12].https://www. bbci.de/competition/iv/.
[22] ALEXANDRE G. MEG and EEG Data Analysis with MNEPython [J].Front Neurosci，2013，7（7）：267.
[23] COHEN M X. A Better Way to Define and Describe Morlet Wavelets for Time-Frequency Analysis [J].Neuroimage，2019，199：81-86.
[24] LE S，GORDON E，GYSELS E. Phase Synchrony Rate for the Recognition of Motor Imagery in Brain-Computer Interface [C]// Advances in Neural Information Processing Systems.Vancouver： DBLP，2005：1265-1272.
[25] BURGES C J C. A Tutorial on Support Vector Machines for Pattern Recognition [J].Data Mining and Knowledge Discovery，1998 （2）：121-167.
[26] RAZA H，CECOTTI H，PRASAD G. A Combination of Transductive and Inductive Learning for Handling Non-Stationarities in Motor Imagery Classification [C]//2006 International Joint Conference on Neural Networks（IJCNN）.IEEE：Vancouver，2016：763-770.
作者简介：张帅（1995—），男，汉族，山西忻州人，硕士研究生在读，研究方向：智能数据分析与应用。