基于黎曼与巴氏距离的脑磁图信号分类方法

发布时间：2018-05-28 09:00

本文选题：脑磁图(MEG) + 分类算法　；参考：《计算机科学与探索》2017年05期

【摘要】：针对人脑对不同视觉目标刺激产生的脑磁图(magnetoencephalography,MEG)信号,提出了一种新型的脑磁图信号分类算法。该算法首先将滤波后的脑磁图信号投影到新的特征空间,然后将脑磁图信号投影后新特征的协方差特征投影到切线空间中,用协方差特征作为信号的特征,进而对样本进行预分类;接着将预分类的样本通过巴氏距离的调整,得到二次标记结果;最后采用黎曼距离对协方差特征矩阵在流形上进行调整,得到最终的分类结果。实验结果表明,该有监督与无监督相结合的算法有助于提高脑磁图信号分类的准确率。
[Abstract]:In view of the magnetoencephalography (MEG) signal generated by the human brain for different visual targets, a new classification algorithm for magnetoencephalography (magnetoencephalography) is proposed, which first projections the filtered brain magnetograph signals to the new feature space, and then projection the covariance features of the new features of the magnetoencephalography signals to the tangent space. The covariance feature is used as the characteristic of the signal, and then the sample is pre classified, and then the two marking results are obtained by adjusting the pre classified samples through the adjustment of the barren distance. Finally, the Riemann distance is used to adjust the covariance feature matrix on the manifold to get the final classification results. The experimental results show that it is supervised and unsupervised. The combined algorithm is helpful to improve the accuracy of classification of magnetoencephalography signals.
【作者单位】：天津大学电子信息工程学院;
【基金】：国家自然科学基金Nos.61372145,61172121,61002030,61002027~~
【分类号】：R338;TP391.41

【相似文献】