摘要
数据集中各数据项间匹配关系异常的样本是难以检测的,针对此问题,文章提出了基于小波分析的异常样本检测与修复方法,该方法利用小波分析的多尺度、局部分析等特性,能有效地实现异常数据样本准确检测和修复.为了实现离散序列小波变换的快速计算,还提出了一种基于Newton-Cores公式修正后的数值积分算法.测试结果表明,该方法切实可行、效果良好、有很强的实用性.*
It is difficult to detect the anomalies of which the relationship among its attributes is very different from one another in a data set. Aiming at this problem, an approach based on wavelet analysis to detect and amend the anomalous samples is proposed in this paper. Taking full advantage of wavelet analysis characteristics of multiple scale and local analysis, this approach is able to detect and amend anomalous samples accurately. To realize the rapid numerical computation of wavelet translation for a discrete sequence, a modified algorithm based on Newton-Cores formula is also proposed. The test results show that the approach is feasible, with good effect and practicality.
出处
《信息与控制》
CSCD
北大核心
2005年第6期676-679,共4页
Information and Control
基金
国家自然科学基金资助项目(50374079)
国家博士点基金资助项目(20030533008)
