The application of neural networks in intrusion detection can overcome some shortcomings of traditional detection technology, such as high false alarm rate and inability to detect variants of known attacks. However, general neural network models can only determine whether the system is attacked, but cannot know what type of attack it is. The multi-neural network model and training method described in this article can not only determine whether it is attacked, but also identify the type of attack. Keywords: neural network, feature extraction, BP algorithm with impulse. With the development of Internet technology and changes in network security, traditional intrusion detection technology is increasingly exposing its limitations and shortcomings. Applying neural networks to intrusion detection systems has many advantages that traditional detection methods cannot match: 1. Neural networks have the ability to generalize and abstract, and have a certain degree of fault tolerance for incomplete input information. 2. Neural networks have a high degree of learning and adaptive capabilities. 3. Neural networks have their inherent parallelism, and each node can be calculated in parallel. This potential high-speed computing capability means that more information can be processed in a very short time. Improve the detection speed. With the development of the Internet environment, intrusion detection technology based on network traffic analysis has become increasingly popular. Cannady (reference [1]) and Mahaffey applied the MLP model and the SOM/MLP hybrid model to the misuse detection model based on network traffic. Lippmann and Cunningham (reference [2]) of MIT proposed the use of a combination of keywords and neural networks for network intrusion detection and conducted relevant research on Telnet service sessions. Lippmann et al. explained the principle of keyword selection and reduced the false alarm rate to about once a day while achieving an 80% detection rate.
You Might Like
Recommended ContentMore
Open source project More
Popular Components
Searched by Users
Just Take a LookMore
Trending Downloads
Trending ArticlesMore