应用于心电图分类的两种数据输入模式的对比

Contrast Research on Two Modes of Inputting Data Applied to Electrocardiography Data Classification

两分类问题是模式识别的重要研究领域,在分类器选定的情况下如何进一步提高分类正确率是一个普遍受到关注的问题。为此,根据心电信号数据有8个导联且反映心电向量在不同方向情况的特点,发现一种不同以往的对训练样本进行运用的方法,并且与已往常用的数据输入模式在应用于心电图信号分类问题上进行了对比实验。新数据输入模式不仅降低了输入空间的维数,而且充分利用了心电信号具有的医学特征,分散了只依靠一个导联作决定的风险和错误因素。实验结果表明这是一个值得运用的数据处理方式。

Two-class classifying is very important in pattern recognition field. After the classifier is determined, more attention is paid to further enhance the accuracy. According to the characteristics of the ECG data, a new mode to use the samples was presented. Contrast experiments were made between this mode and other modes. Experiments show that by this new training mode, not only the accuracy is improved, but also the performance of this algorithm is of high stability. The new mode decreases the sample＇s dimensions in the feature space, and considers the eight conductors comprehensively as well＇ so as to decentralize the risk and mistaken factors, Experiment results show that this data-inputting mode is worth employing.