基于相关向量机的神经活动分类及译码

Nerval Activity Classification and Decoding Based on Relevance Vector Machine

脑机接口研究受到越来越多学者的关注,其中对神经活动的分类和译码是研究的重要方面。利用相关向量机的方法对来自脑皮层的一部分运动神经元的激发率进行分类,识别其神经状态,在此基础上利用激发率进行译码,判断其运动轨迹。实验证明,相关向量机能够较好地进行神经活动的分类和译码,并且拥有比支持向量机和信息向量机更好的性能。

The research of brain-computer interface attracts more and more interests,especially the classification and decoding of nerval activity is most important.This paper uses relevance vector machine algorithm to classify the firing rates from small populations of neurons in primary motor cortex.It uses the output of classifier to recursively infer nerval state and hand kinematics conditioned on neural firing rates.Experiments show that the relevance vector machine algorithm is suited for the classification and decoding of nerval activity, and the performance of relevance vector machine is better than the popular support vector machine and information vector machine.