RBF神经网络抗体群用于入侵检测的研究  

Study of intrusion detection based on RBF neural network antibody group

作　　者：何夏冰[1] 薛波[1] 陈崚[1,2] 徐永安[1] 

机构地区：[1]扬州大学信息工程学院,江苏扬州225009 [2]南京大学软件新技术国家重点实验室,江苏南京210093

出　　处：《计算机工程与设计》2011年第10期3254-3257,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(61070047)

摘　　要：为了进一步模仿生物免疫系统,提升入侵检测系统的性能,提出了具有抗原针对性的抗体生成算法和具有抗体针对性的抗原检测算法。对不同类型的已知抗原,生成相应类型的抗体,对不同类型的外来抗原,用相应类型的抗体检测,提高了对生物免疫系统的模仿程度;通过免疫应答、克隆选择和疫苗注射来训练生成神经网络抗体群。仿真实验结果表明,该算法具有很高的精度和很强的自适应能力。To further mimic biological immune system, improve the performance of intrusion detection system, a antigen specificantibody production algorithm and a antibody specific antigen detection algorithm are proposed. Known for different types of antigens to generate appropriate types of antibodies, different types of foreign antigens, with corresponding types of antibodies detection of biological immune system, the degree of imitation is improved. The neural network antibody group is generated through immune response, clonal selection and vaccination. Simulation results show that the algorithm has good precision and strong adaptability.

关 键 词：入侵检测 人工免疫 径向基函数 神经网络 抗体群 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]