基于贝叶斯和F-measure感知机的分类器设计

Design of classifier based on Bayes and F-measure perceptron

当样本特征向量交织时,分类很容易出错。为解决这个问题,提出一种基于Bayes和F-measure的分类器算法。采用替代方法评估分类器的性能正受到关注,特别是对于不平衡的问题。该算法利用F-measure分析不平衡数据的分类准确度,将类概率密度函数引入判据,并采用梯度下降法得到准则函数。文中将所提出的方法与传统方法进行比较,实验结果表明,该方法能够有效提高识别的准确率和精确度。

A classifier algorithm based on Bayes and F-measure is proposed to solve the problem that the classification is prone to error when the sample feature vectors are intertwined.The alternative methods used for the performance evaluation of classifiers are receiving increasing attention,especially for unbalanced data classification.The algorithm is used to analyze the classification accuracy of the unbalanced data by means of F-measure.The probability density function is introduced into the criterion,and the gradient descent method is used to obtain criterion function.The proposed method is compared with the traditional ones,in this paper.The experimental results show that the proposed method can effectively improve the accuracy and precision of recognition.