蚁群聚类算法在隐写分析中的应用

为解决隐写分析中富模型的特征维数较高、冗余较大、不便于高效分类的问题，提出了一种基于蚁群聚类算法的降维方法。首先利用蚁群聚类算法求解特征簇的簇中心，然后把簇中心作为新的特征，提取新特征的有效部分用集成特征进行分类。实验结果表明，利用蚁群聚类算法对高维特征进行降维，可以有效去除冗余特征，提升特征的分类效果。...

Full description

Saved in:
Bibliographic Details
Published in计算机应用研究 Vol. 32; no. 9; pp. 2803 - 2805
Main Author 马占山 张敏情 钮可 苏光伟
Format Journal Article
LanguageChinese
Published 武警工程大学 电子技术系 网络与信息安全武警部队重点实验室,西安,710086 2015
Subjects
富模型
蚁群算法
隐写分析
集成分类
富模型
蚁群算法
rich model
ensemble classification
ant colony algorithm
集成分类
隐写分析
steganalysis
Online AccessGet full text
ISSN1001-3695
DOI10.3969/j.issn.1001-3695.2015.09.056

Cover

More Information
Summary:为解决隐写分析中富模型的特征维数较高、冗余较大、不便于高效分类的问题，提出了一种基于蚁群聚类算法的降维方法。首先利用蚁群聚类算法求解特征簇的簇中心，然后把簇中心作为新的特征，提取新特征的有效部分用集成特征进行分类。实验结果表明，利用蚁群聚类算法对高维特征进行降维，可以有效去除冗余特征，提升特征的分类效果。
Bibliography:This paper proposed a means of dimension reduction which was based on ant colony clustering algorithm to solve the problem of steganlysis that the features had higher dimension and bigger redundancy in rich model ,which was not convenient to classify efficiently. First, this method collected the cluster center from the features of the cluster by ant colony clustering al- gorithm, Then, it regarded the cluster center as a new feature to extract the effective part from the new features and used the ensemble classifier to make classification. It turns out that reduce dimension with ant colony clustering algorithmcan remove some redundant features effectively and improve the effect of features.
51-1196/TP
Ma Zhanshan, Zhang Minqing, Niu Ke, Su Guangwei （ Key Laboratory of Network ＆ Information Security Under the Armed Police Force, Dept. of Electronic Technology, Engineering University of Chinese Armed Police Force, Xi ＇ an 710086, China）
steganalysis; rich model; ensemble classification; ant colony algorithm
ISSN:1001-3695
DOI:10.3969/j.issn.1001-3695.2015.09.056