一种基于ART-2改进后的网格入侵检测模型

Improved ART-2-based Grid Intrusion Detection Model

应用神经网络中的ART-2理论(adaptive resonance theory自适应共振理论),在传统ART-2模型的基础上增加了伴随神经元和重置系统B,解决入侵检测系统中可能出现的对渐变过程不敏感从而导致的预分类不完全的问题,通过与基于传统ART-2的入侵检测模型及基于朴素贝叶斯的入侵检测模型的对比,发现改进后的ART-2神经网络打破了传统ART-2对渐变过程不敏感的局限性,使得新模型能够分辨渐变过程,提高了预分类的能力。

The neural network ART-2 theory （adaptive resonance theory） was applied, based on the traditional ART-2, the adjoint neuron and reset system B were increased to have solved the problem of the incomplete pre-classification caused by the insensitivity to the process of gradual change in the intrusion detection system. By comparison with the intrusion detection model based on the traditional ART-2 and the intgusion detection model based on Naive Bayes, the results expatiated the improved ART-2 neural network broke the limitation of traditional ART-2 insensitive to the process of gradual change and improved the capacity of the pre-classification.