论文部分内容阅读
本文提出了一个基于机器学习的新方法,用以从单实验脑电信号中估计事件相关电位.为估计事件相关电位,首先构建了一个运用分类方法的基本框架和以此框架为基础的优化模型.在此基础上,引入logistic回归来实例化这个优化模型,并推导出SingleTrialEM算法.只要在使用之前训练得到一个logistic模型,SingleTrialEM算法就能够从单实验脑电信号中估计事件相关电位.模拟测试表明,本文的方法是正确稳定的,明显优于Woody过滤器方法.认知测试的结果与认知科学的各项结论一致.
In this paper, a new method based on machine learning is proposed to estimate the event-related potential from a single experimental EEG. To estimate the event-related potential, a basic framework using classification methods and an optimization model based on this framework Based on this, logistic regression is introduced to instantiate this optimization model and derive the SingleTrialEM algorithm.As long as a logistic model is trained prior to use, the SingleTrialEM algorithm can estimate the event-related potential from a single experimental EEG. The results show that the method in this paper is correct and stable, which is obviously superior to Woody’s filter method. The results of cognitive tests are consistent with those of cognitive science.