项目名称: 可拓支持向量机理论、方法与应用研究
项目编号: No.61472390
项目类型: 面上项目
立项/批准年度: 2015
项目学科: 其他
项目作者: 田英杰
作者单位: 中国科学院大学
项目金额: 80万元
中文摘要: 本项目将从可拓学与最优化理论的角度,研究基于可拓学的支持向量机理论、方法与应用,并建立可拓支持向量机理论新体系,这一研究具有重要的理论意义与应用价值。我们的研究着重两个方面:一是可拓支持向量机理论与方法研究。我们以支持向量机为基础,结合可拓学的基本理论,以最优化与机器学习理论为工具,构建求解数据挖掘问题的全新支持向量机模型,该模型考虑到了大量问题中的事物具有某种性质的程度是变化的,从而具有处理矛盾问题的能力;同时将在该模型基础上,拓展其到诸多实际问题中,建立完善的可拓支持向量机的算法体系;进一步,研究基于知识库的可拓支持向量机,对已有的专家经验或知识进行可拓变换,并与可拓支持向量机结合,以提高数据挖掘的能力。二是基于可拓支持向量机的应用研究。我们将以某商业银行的数据为基础,构建相应的可拓风险控制与管理模型,挖掘有用的可拓知识。
中文关键词: 数据挖掘;支持向量机;可拓学;最优化;机器学习
英文摘要: This research proposal focuses on the Extension Support Vector Machine (ESVM) Theory , Methods and Applications from the Extenics and Optimiazion points of the view, in order to establish the new theoretical system of ESVM, which has the important theoretical significance and application value. We will research on two important parts: The first is the ESVM theory and algorithms research. We take the standard SVM as the basis, Optimizaiton and Machine Learning therory as the tools, then combine the basic theory of Extenics to construct the new SVM. This brand model takes into account that the objects in lots of problems have the varing property, therefore ESVM has the ability to deal with the contradiction. At the same time, we will extend ESVM to various learning problems to bulid the complete system of ESVM.Furthermore, we will research the ESVM based on knowledge sets.Combining ESVM with the existing experts experience or knowledge is to enhance the ability of data mining. The second is the application research of ESVM. Based on the data of some commercial bank in China, we will construct the corresponding extension risk control and management model to mine useful extension knowledge.
英文关键词: Data Mining;Support Vector Machine;Extensics;Optimization;Machine Learning