基于神经网络的入侵识别方法

A Neural Network-based Intrusion Recognition Technique

分析了正常和异常的网络行为模式所带来的网络传输数据包的不同特征，提出了 基于神经网络的入侵识别方法。采用带偏差单元的回归BP算法使神经网络收敛性能良好， 并 对端口扫描类型的攻击进行了仿真识别。

By analyzing the different characteristics of datagram of normal and abnormal network usage patterns, an intrusion recognition technique is present e d with neural network in this paper. Better performance can be gained by BP al go rithm with bias cell and internal recurrent. The result of simulation of port sc anner is also given.