论文部分内容阅读
快速有效地提取和比较不同神经信号中所含信息的相同和区别之处是研究者关注的问题。本文介绍了一种基于小波变换的时频建模方法。该方法用少量的半椭球模型来表征神经信号在时域和频域上的变化,克服了传统时频分析中背景干扰大、参数多的缺陷。将该方法应用于嗅球场电位的研究,与支持向量机(SVM)算法相结合,可初步实现气味的分类。
It is an issue of interest to researchers to quickly and efficiently extract and compare the similarities and differences between the information contained in different neural signals. This paper introduces a time-frequency modeling method based on wavelet transform. This method uses a small amount of semi-ellipsoid model to characterize the changes of neural signals in the time and frequency domains, and overcomes the shortcomings of the traditional time-frequency analysis such as large background interference and many parameters. This method is applied to the study of the potential of the olfactory bulb field. Combined with support vector machine (SVM) algorithm, odor classification can be initially achieved.