摘要
通过将主成分分析(PCA)方法集成到树突细胞算法(DCA)的数据预处理过程中,提出一种可实现信号自动提取,并根据数据及时更新信号提取方案的PCA-DCA算法。利用PCA对样本高维数据进行降维处理,根据贡献度提取输入信号,通过改进的DCA算法对抗原进行分类与检测;通过监测记录的高维数据,不断更新主成分特征向量,保证降维后数据的有效性。通过KDD CUP99数据集对该算法进行验证,相较传统DCA算法,该算法实现了对多维数据的降维处理及对DCA输入信号的自动提取,对异常数据更加敏感,检测效率更高。
By integrating the principal component analysis method to the DCA's data pre-processing phase,an improved DCA with the signal presenting and updating automatically was proposed.The principal components analysis was used for the dimensionality reduction of high-dimensional data,and the DCA's input signal was presented according to the contribution,and these antigens were detected and classified using the improved DCA.The monitoring of high-dimensional data was tracked,principal component eigenvectors were updated constantly,to ensure the validity of dimensionality reduction.The dataset of KDD CUP99 was used to test this algorithm.Compared to traditional DCA,the dimensionality reduction and automatic pre-processing of DCA's input signal were achieved.Results show that the method is more sensitive to abnormity,and gets more effective detection rate.
出处
《计算机工程与设计》
北大核心
2017年第6期1414-1417,1472,共5页
Computer Engineering and Design
基金
军内科研预研基金项目
关键词
树突细胞算法
主成分分析
入侵检测
人工免疫
数据预处理
dendritic cell algorithm
principal component analysis
intrusion detection
artificial immune
data pre-processing
