项目名称: 基于n阶引力场理论的数据分类研究
项目编号: No.61273290
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 许华荣
作者单位: 厦门理工学院
项目金额: 83万元
中文摘要: 数据分类是模式识别和机器学习领域中的核心问题之一,具有重要的理论研究意义和应用价值。随着数据采集、存储、传输手段的不断提高以及数据量的不断增大,如何从海量数据中提取出数据的本质属性,实现准确的数据分类成为一项重要的研究课题。n阶引力场本质上是物理学中万有引力场的n阶泛化场,它可以有效描述引力场中粒子之间的相互作用关系。本项目拟运用n阶引力场理论来研究海量数据的自动分类问题,主要的研究内容包括:(1)n阶引力场中粒子质量的有效表达方式;(2)n阶引力场中斥力的定义与平衡状态的计算方法;(3)基于n阶引力场的分类器构造方法;(4)n阶引力场中参数n的自动学习方法。本项目的研究可望为数据有效分类提供一种新的思路和途径。
中文关键词: n 阶引力场;数据分类;有效质量;平衡状态;
英文摘要: Data classification is one of the key problems in pattern recognition and machine learning, which has great theoretical significance and practical value. With the improvements of data acquisition, data storage, data transmission as well as with the continuous increase of the data volume, how to extract the intrinsic attributes from the data and achieve accurate classification becomes an important research topic. The nth power gravitational field is essentially a generalized version of the Newton's universal gravitation field in physics, which is able to describe the interactive relationships among particles in the field effectively. This project employs the nth power gravitational field theory for data classification, and the key issues include:(1) An effective representation for the particle mass in the nth power gravitational field; (2) A definition on the repulsive force, and methodology to compute the equilibrium of the nth power gravitational field; (3) Algorithms for constructing classifiers based on the nth power gravitational field theory; (4) An algorithm for automatically learning the parameter n in the nth power gravitational field. The outcome of the present proposal could provide a novel approach and methodology for data classification.
英文关键词: the nth power gravitational field;data classfication;effective mass;equilibrium;