使用广义Volterra核模型预测肌电图
Predicting EMG with generalized Volterra kernel model.
作者信息
Song Dong, Hendrickson Phillip, Marmarelis Vasilis Z, Aguayo Jose, He Jiping, Loeb Gerald E, Berger Theodore W
机构信息
Department of Biomedical Engineering, Center for Neural Engineering, University of Southern California, Los Angeles, CA 90089, USA.
出版信息
Annu Int Conf IEEE Eng Med Biol Soc. 2008;2008:201-4. doi: 10.1109/IEMBS.2008.4649125.
Generalized Volterra kernel model (GVM) is developed in spirits of the generalized linear model (GLM) and used to predict EMG signals based on M1 cortical spike trains during a prehension task. The GVM for EMG consists of a cascade of a multiple-input-single-output Volterra kernel model (VM) and an exponential activation function. Without loss of generality, the exponential activation function constrains the unbounded VM output within the positive range, which fully covers the dynamic range of the rectified EMG signals. Results show that GVMs are more accurate than the VMs due to this asymptotic property.
广义Volterra核模型（GVM）是在广义线性模型（GLM）的基础上发展而来的，用于在抓握任务期间基于M1皮质尖峰序列预测肌电信号。用于肌电的GVM由一个多输入单输出Volterra核模型（VM）和一个指数激活函数级联组成。不失一般性，指数激活函数将无界的VM输出限制在正范围内，该范围完全覆盖了经整流的肌电信号的动态范围。结果表明，由于这种渐近特性，GVM比VM更准确。
Zotero 插件