一种基于迁移softmax的文本分类方法

河南科技大学本科毕业设计（论文）A text classification method based on migration softmaxABSTRACTIn the process of the increasing improvement of Internet informationtechnology,more and more data and information can be received by people.Among them,text information has advantages of occupying less network contentthan voice information and image information,and is convenient for uploadingand downloading,and content is easy to read.This makes most of the resourcesand information in the network in the form of text.In order to more effectively classify texts,the main research goals of thispaper are proposed,focusing on solving the problem of text informationextraction.In large amounts of text information data,useful text information isobtained by filtering out invalid text.Based on the existing machine learning textclassification system,it can be learned that the implementation of textclassification is based on the corresponding algorithm rules,or with the help ofmathematical modeling ideas.Therefore,the accuracy of text classification isclosely related to the classification model.This topic first introduces the background of text classification research in theresearch,and then analyzes its source,and learns the corresponding knowledgeto lay the theoretical foundation for the research of this question.This article also introduces the relevant knowledge of transfer learning,theprinciple of softmax regression model,by introducing the characteristics of theabove model,and combining with the reasonable application of computertechnology,respectively,automatically classify the text content of two groups ofsimilar characteristics,and plan separately to the set Good data set program.Andgroup them into the source domain dataset and the target domain dataset;then,