脑机接口中基于MRP的半监督判决空间模式法

Semi-Supervised Discriminative Spatial Patterns Based on MRP for Brain-Computer Interfaces

在脑-机接口研究中,如果训练样本少,判决空间模式法不能很好地提取运动相关电位特征。为此,文中在半监督框架下,采用自训练方法,引入分类置信度高的无标记样本,迭代学习MRP的空间判决模式。实验结果验证了所提算法的有效性。

In the study of brain-computer interface,if the number of training samples is small,the features of movement related potentials can not be well extracted by discriminative spatial pattern algorithm.Thus in this paper,semi-supervised self-training scheme is employed to induce the unlabelled samples with high confidences and learn the discriminative patterns of MRPs iteratively.The results of experiments demonstrate the effectiveness of the proposed algorithm.