摘要
离群点检测在欺诈检测、网络鲁棒性分析和入侵检测等领域有着重要的应用。Aggarwal和Yu提出的基于子空间投影和遗传算法(GA)的离群点检测方法是处理高维数据的一个有效方法。由于该算法的交叉重组过程采用贪心策略选择子串,并且随着变异概率的改变可能导致发现不了一些有意义的离群数据。文中对该算法的交叉过程和变异过程进行改进,提出一种改进的算法,提高了检测的精度并且不受变异概率改变的影响。
The outlier detection problem has important applications in the field of fraud detection, network robusmess analysis and intrusion detection. Aggarwal and Yu＇s recent projection of space-based and genetic algorithm （GA） of the outlier detection of high-dimensional data processing is an effective method,because the algorithm is cross-restructuring process by greedy string of strategic options, and mutation in the course of this mutation probability that the change may not lead to some outlier. Presents an improved method, through the crossover and the variability of due process of improvement, improve the accuracy of the test and is not subject to mutation probability of impact.
出处
《计算机技术与发展》
2009年第3期141-143,147,共4页
Computer Technology and Development
基金
安徽省高等学校省级自然科学研究项目(kj2008B092)
关键词
离群点检测
高维数据
遗传算法
交叉
变异
outlier detection
high-dimensional data
genetic algorithm
crossover
mutation
