通过叠加少量正弦波实现单词的脑电波表示。
Brain-wave representation of words by superposition of a few sine waves.
作者信息
Suppes P, Han B
机构信息
Center for the Study of Language and Information and Department of Management Science and Engineering, Stanford University, Stanford, CA 94305, USA.
出版信息
Proc Natl Acad Sci U S A. 2000 Jul 18;97(15):8738-43. doi: 10.1073/pnas.140228397.
Data from three previous experiments were analyzed to test the hypothesis that brain waves of spoken or written words can be represented by the superposition of a few sine waves. First, we averaged the data over trials and a set of subjects, and, in one case, over experimental conditions as well. Next we applied a Fourier transform to the averaged data and selected those frequencies with high energy, in no case more than nine in number. The superpositions of these selected sine waves were taken as prototypes. The averaged unfiltered data were the test samples. The prototypes were used to classify the test samples according to a least-squares criterion of fit. The results were seven of seven correct classifications for the first experiment using only three frequencies, six of eight for the second experiment using nine frequencies, and eight of eight for the third experiment using five frequencies.
分析了来自之前三个实验的数据，以检验口语或书面文字的脑电波可以由少数正弦波叠加表示这一假设。首先，我们对多个试次和一组受试者的数据进行平均，在一种情况下，还对实验条件进行了平均。接下来，我们对平均后的数据应用傅里叶变换，并选择能量高的那些频率，数量上在任何情况下都不超过九个。这些选定正弦波的叠加被用作原型。未经滤波的平均数据是测试样本。根据最小二乘拟合标准，使用原型对测试样本进行分类。结果是，第一个实验仅使用三个频率，七次分类全部正确；第二个实验使用九个频率，八次中有六次正确；第三个实验使用五个频率，八次分类全部正确。
Zotero 插件