基于拟牛顿算法优化神经网络的入侵检测研究

Research on intrusion detection based on the Quasi-Newton algorithm in neural networks

当前的入侵检测技术主要有基于规则的误用检测和基于统计的异常检测。提出了一种基于拟牛顿算法优化神经网络的入侵检测方法,与传统算法相比,该方法具有收敛快,检测率高等优点。

The current intrusion detection techniques mainly include role-based misuse detection and statistics-based anomaly detection. The Quasi-Newton algorithm is taken to optimize traditional BP neural network in this paper. Compared with the traditional BP algorithm, the new method has higher speed in constringency and more precise in detection.