Abstract: By analyzing the reasons for the low detection rate of the existing real-value negative selection algorithms, an improved negative selection algorithm is proposed with the identification of boundary samples to improve the coverage of holes. Detailed realization and advantages of the algorithm are given. The experiments of synthetic 2DSyntheticData and real biomedical data sets are done to test the algorithm. Experimental results show that the algorithm has higher detection rate and needs less detector numbers. It has optimum overall performance.

Key words: artificial immune system, negative selection algorithm, anomaly detection, real-value, data set

摘要： 通过分析已有实值负选择算法检测率不高的原因，提出一种通过鉴别边界自体样本的改进负选择算法，以提高对检测黑洞的覆盖率。给出算法的改进思想、具体实现过程及优势分析。采用人工合成数据集2DSyntheticData和实际Biomedical数据集对算法进行验证。实验结果表明，该算法检测率较高，所需的检测器数量较少，综合性能较优。

关键词: 人工免疫系统, 负选择算法, 异常检测, 实值, 数据集

CLC Number: 

• TP309