| 一类分类马氏椭球学习机的直推式学习 |
| |
| 引用本文： | 李建民,李永新,薛贞霞.一类分类马氏椭球学习机的直推式学习[J].计算机仿真,2009,26(8):84-88. |
| |
| 作者姓名： | 李建民 李永新 薛贞霞 |
| |
| 作者单位： | 1. 平顶山学院数学系,河南,平顶山,467002
2. 河南科技大学数学系,河南,洛阳,471003;西安电子科技大学应用数学系,陕西,西安,710071 |
| |
| 基金项目： | 国家自然科学基金项目 |
| |
| 摘 要： | 针对一类分类马氏椭球学习机当训练样本点比较少而待分类的样本点比较多时,分类精度不高,系统适应性不强的问题,提出直推式一类分类马氏椭球学习机.为解决上述问题,在训练过程中利用已知的训练样本点和待分类的样本点的信息,将待分类样本点逐渐加人到学习机中,并能有效地利用历史训练结果对其进行识别和分类,具有增量学习的特点.与一类分类马氏椭球学习机相比,方法能在很小的训练样本集规模下提高学习机的分类精度,从而使系统的适应性更好.仿真数据和真实数据的实验表明直推式一类分类马氏椭球学习机能大幅度地提高学习的精度.
|
| 关 键 词： | 模式识别 直推式学习 超椭球 |
Transductive Learning Mahalanobis Ellipsoidal Learning Machine for One Class Classification |
| |
| LI Jian-min,LI Yong-xin,XUE Zhen-xia.Transductive Learning Mahalanobis Ellipsoidal Learning Machine for One Class Classification[J].Computer Simulation,2009,26(8):84-88. |
| |
| Authors: | LI Jian-min LI Yong-xin XUE Zhen-xia |
| |
| Affiliation: | 1.Department of Mathematics;Pingdingshan University;Pingdingshan Henan 467002;China;2.Department of Mathematics;Henan University of Science and Technology;Luoyang Henan 471003;3.Department of Applied Mathematics;Xidian University;Xi'an Shanxi 710071 China |
| |
| Abstract: | In Mahalanobis ellipsoidal learning machine for one class classification(MELMOCC),the classification accuracy is not high and the adaptability of the system is not good when the training samples are fewer and the samples to be classified are much more.A transductive Mahalanobis ellipsoidal learning machine for one Class Classification(TMELMOCC) is proposed,which utilizes the information of both training samples and samples to be classified during the training process,adds gradually samples to be classified ... |
| |
| Keywords: | Pattern recognition Transductive learning Hyper-ellipsoid |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
