| Author Name | Affiliation | | ZHANG Feng-bin | Dept.of Computer Science & Technology,Harbin University of Science & Technology,Harbin 150080,China Dept.of Computer Science & Technology,Harbin Institute of Technology,Harbin 150001,China | | WANG Da-wei | Dept.of Computer Science & Technology,Harbin University of Science & Technology,Harbin 150080,China | | WANG Sheng-wen | Dept.of Computer Science & Technology,Tsinghua University,Beijing 100084,China |
|
| Abstract: |
| This paper proposes a negative selection with neighborhood representation named as neighborhood negative selection algorithm.This algorithm employs a new representation method which uses the fully adjacent but mutually disjoint neighborhoods to present the self samples and detectors.After normalizing the normal samples into neighborhood shape space,the algorithm uses a special matching rule similar as Hamming distance to train mature detectors at the training stage and detect anomaly at the detection stage.The neighborhood negative selection algorithm is tested using KDD CUP 1999 dataset.Experimental results show that the algorithm can prevent the negative effect of the dimension of shape space,and provide a more accuracy and stable detection performance. |
| Key words: artificial immune negative selection neighborhood matching rule |
| DOI:10.11916/j.issn.1005-9113.2011.03.014 |
| Clc Number:TP301.6 |
| Fund: Sponsored by the National Natural Science Foundation of China (Grant No. 60671049);the Subject Chief Foundation of Harbin (Grant No.2003AFXXJ013);the Education Department Research Foundation of Heilongjiang Province(Grant No. 10541044 and 1151G012);the Postdoctoral Science-research Developmental Foundation of Heilongjiang Province(Grant No. LBH-Q09075) |