基于双隐层极限学习机的模糊XML文档分类  

Classification of fuzzy XML documents based on double hidden layer ELM

作　　者：赵震[1,2] 马宗民[1] 张富[1] 林晓庆[1] ZHAO Zhen;MA Zongmin;ZHANG Fu;LIN Xiaoqing(School of Computer Science and Engineering, Northeastern University, Shenyang 110819, China;College of Information Science and Technology, Bohai University, Jinzhou, Liaoning 121013, China)

机构地区：[1]东北大学计算机科学与工程学院,沈阳110819 [2]渤海大学信息科学与技术学院,辽宁锦州121013

出　　处：《计算机工程与应用》2017年第4期19-24,共6页Computer Engineering and Applications

基　　金：国家自然科学基金(No.61370075;No.61073139;No.61202260);教育部新世纪优秀人才支持计划项目(No.NCET-05-0288)

摘　　要：随着大数据时代的到来,对异构和分布式的模糊XML数据管理显得越来越重要。在模糊XML数据的管理中,模糊XML文档的分类是关键问题。针对模糊XML文档的分类,提出采用双隐层极限学习机模型来实现模糊XML文档自动分类。这个模型可以分为两个部分:第一层采用极限学习机提取模糊XML文档的相应特征,第二层利用核极限学习机根据这些特征进行最终的模糊XML文档分类。通过实验验证了所提方法的性能优势。首先对主要的调节参数包括隐藏层节点的数目L,常量C和核参数γ进行了研究,接下来的对比实验说明提出的基于双隐层ELM(Extreme Learning Machine)的方法相较于传统单隐层ELM以及SVM(Support Vector Machine)方法,分类精度得到较大提高,训练时间进一步缩减。With the arrival of the era of big data, the management of distributed and heterogeneous fuzzy XML data is also becoming more and more important. In the management of fuzzy XML data, the classification of fuzzy XML documents is the key problem. In order to study the classification for fuzzy XML documents, in this paper, a new ELM-based double hidden layer framework is proposed. The proposed architecture is divided into two main components:the feature extraction of fuzzy XML documents are performed using Extreme Learning Machine in first layer, and then use these characteristics to classify the fuzzy XML documents by KELM Kernel Extreme Learning Machine in second layer. Finally, the performance advantages of the proposed method are verified by experiments. Firstly, the parameters including the number of hidden neuron, and the constant parameter C and kernel parameter γ are investigated in detail. Compared with the traditional single hidden layer ELM(Extreme Learning Machine)and SVM(Support Vector Machine)method, the classification accuracy has been greatly improved and the training time has been decreased by approach based on the double hidden layer ELM proposed in this paper.

关 键 词：模糊 XML文档 分类 特征提取 极限学习机 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]