结构大间隔单类分类器

山东大学学报(工学版) ›› 2010, Vol. 40 ›› Issue (3): 6-12.

结构大间隔单类分类器

冯爱民¹,刘学军¹,陈斌²

1. 南京航空航天大学信息科学与技术学院,江苏南京 210016; 2. 扬州大学信息工程学院, 江苏扬州 225009

收稿日期:2010-01-26 出版日期:2010-06-16 发布日期:2010-01-26
作者简介:冯爱民(1971-),女,河南焦作人,副教授,主要研究方向为机器学习、模式识别及异常检测.E-mail: amfeng@nuaa.edu.cn
基金资助:
国家自然科学基金资助项目(60703016,60903130)

Structure large margin one-class classifier

FENG Ai-min¹, LIU Xue-jun¹, CHEN Bin²

1. Information Science ＆ Technology College, Nanjing University of Aeronautics ＆ Astronautics, Nanjing 210016, China;
2. College of Information Engineering, Yangzhou University, Yangzhou 225009, China

Received:2010-01-26 Online:2010-06-16 Published:2010-01-26

摘要/Abstract

摘要：

目标数据呈簇分布、基于超平面的单类分类器要求嵌入结构信息时,必须分别考虑各簇数据对超平面的影响,为此,提出可用于簇分布的结构大间隔单类分类器(structural large margin one-class classifier,SLMOCC)。该算法通过分别约束各簇数据到超平面的马氏距离,并最大化最小马氏间隔,保证目标数据落入正半空间的同时,充分利用数据的簇结构信息,通过序列二次锥规划优化方法线性搜索到最优超平面。为捕捉数据簇结构,SLMOCC采用凝聚型层次聚类并借助拐点确定聚类数目,最后通过人工数据和UCI数据集与相关算法比较,验证了SLMOCC的有效性。

关键词: 单类分类器, 簇结构信息, 马氏距离

Abstract:

In one-class classifier(OCC) design, considering the structure of the target data is a possible way to improve the generalization ability of the model. However, while the targets follow multicluster distributions, it is more reasonable to consider each cluster’s structure individually rather than just to treat all of them as a whole. The novel algorithm structure large margin OCC(SLMOCC) fulfills the above strategy by restricting each data’s Mahalanobis distance to the hyperplane. Through maximizing the minimum Mahalanobis margin, SLMOCC is able to find the more reasonable optimal hyperplane attributed to its finer cluster granularity description compared with other alternatives. As for extracting the underlying data structure, this work adopts the Ward’s agglomerative hierarchical clustering on input data or data mapping in kernel space. Experimental results on toy data and UCI benchmark datasets have shown that SLMOCC outperforms other structural OCCs.

Key words: one-class classifier;cluster structure information, Mahalanobis distance

冯爱民1,刘学军1,陈斌2. 结构大间隔单类分类器[J]. 山东大学学报(工学版), 2010, 40(3): 6-12.

FENG Ai-min1, LIU Xue-jun1, CHEN Bin2. Structure large margin one-class classifier[J]. JOURNAL OF SHANDONG UNIVERSITY (ENGINEERING SCIENCE), 2010, 40(3): 6-12.

参考文献

多维度评价

Viewed

Full text

Abstract

Cited

Shared

Discussed

[1]	姚宇,冯健,张化光,韩克镇. 一种基于椭球体支持向量描述的异常检测方法[J]. 山东大学学报(工学版), 2017, 47(5): 195-202.
[2]	陶志伟,张莉. 基于马氏距离的分段矢量量化时间序列分类[J]. 山东大学学报(工学版), 2016, 46(3): 51-57.
[3]	赵加敏,冯爱民*,刘学军. 局部密度嵌入的结构单类支持向量机[J]. 山东大学学报(工学版), 2012, 42(4): 13-18.
[4]	张思懿1,2,王士同1*. 核化空间深度间距的特征提取方法[J]. 山东大学学报(工学版), 2012, 42(3): 45-51.