论文部分内容阅读
文章在简单介绍了入侵检测技术之后,在前人工作的基础上提出了一种在异常检测中用神经网络构建程序行为的特征轮廓的思想。文中给出了神经网络算法的选择和应用神经网络的两种网络设计方案,并对它们进行了比较。实验表明在异常检测中用神经网络构建程序行为的特征轮廓,能够大大提高检测系统对偶然事件和入侵变异的自适应性,特别是带有反馈的回归神经网络能更充分地利用数据信息,在保持系统的虚警率不变的情况下使检测率也有所提高。
After introducing the technology of intrusion detection briefly, this paper puts forward a thought of using neural network to construct the outline of program behavior in anomaly detection based on the previous work. In this paper, two kinds of network design schemes of neural network algorithm selection and neural network application are given, and compared with each other. Experiments show that using neural network to construct the feature outline of program behavior in anomaly detection can greatly improve the adaptability of the detection system to accidental events and invading mutation. Especially, the regression neural network with feedback can make full use of data information. Maintaining the false alarm rate unchanged, the detection rate also increased.