Abstract: This paper uses Informative Vector Machine(IVM) algorithm to classify the firing rates from small populations of neurons in primary motor cortex. It uses the output of classifier to recursively infer neural state and hand kinematics conditioned on neural firing rates. Experiments are done to compare IVM with Support Vector Machine(SVM) and Relevance Vector Machine(RVM), and the results demonstrate that the classification and decoding performance of neural activity of IVM is best, and its run time is least.

Key words: Informative Vector Machine(IVM), sparse Gaussian process, neural activity classification and decoding, Support Vector Machine (SVM), Relevance Vector Machine(RVM)

摘要： 利用信息向量机(IVM)算法对来自脑皮层的少量运动神经元激发率进行分类，识别出神经状态，用该神经状态指导神经元激发率的译码。将IVM算法与支持向量机算法、相关向量机算法进行实验比较，结果证明，IVM算法的神经活动分类和译码性能最优，运行时间最短。

关键词: 信息向量机, 稀疏高斯过程, 神经活动分类和译码, 支持向量机, 相关向量机

CLC Number: 

• TP18