Abstract: A multi-detector IDS model based on computer cluster is put forward. In this system every detector comprises several PCs that can improve the response speed of the system’s. In order to improve the prediction and generalization ability of the detector, twice PSO algorithm is proposed to construct neural network ensemble, which is designed in PVM parallel environment.

Key words: intrusion detection system(IDS), neural network ensemble, selective ensemble, PSO, parallel computation

摘要： 在系统结构上提出了一种多检测器并行的智能机群入侵检测系统模型，系统中每一个检测器是一个神经网络集成分类检测器，由多个PC组成，以提高系统响应速度。采用两次粒子群优化算法选择性集成神经网络集，提高了神经网络集成检测器的预测精度。程序设计采用PVM并行方式实现。

关键词: 入侵检测系统, 神经网络集成, 选择性集成, 粒子群优化, 并行计算

CLC Number: 

• TP393