Intrusion detection method based on multistage fuzzy neural network

Abstract

The invention discloses an intrusion detection method based on a multistage fuzzy neural network. The intrusion detection method comprises the steps of network data preprocessing, super box creation and adjustment, overlapped area processing and classification result display. An improved fuzzy minimum-maximum neural network algorithm is taken as basis. An international standard data set KDD CUP99network connection data set is taken as an example. An experiment selects corrected KDDCUP99 dataset given in 2007 and used for comparison and verification of algorithm performance. The method comprises the following steps: prior to the experiment, preprocessing experimental data, constructing a hyperbox according to the preprocessed network connection data, adjusting the hyperbox, processing an overlapped area, and finally realizing classification of test network connection data, thereby judging whether the current network connection is an attack connection and a specific attack type. The intrusion detection method based on the multistage fuzzy neural network greatly improves the network connection classification speed and classification accuracy, and improves the defects of slow classification and high false alarm rate of a traditional intrusion detection system to a certain extent.

Description

technical field [0001] The method relates to the field of network intrusion detection systems, in particular to an intrusion detection method based on a multi-level fuzzy neural network. Background technique [0002] With the rapid development of Internet technology, computer network security issues are always threatening people's daily life and work, and have attracted people's attention. How to identify various network attacks, especially unforeseen attacks, is an inevitable key technical issue. Network intrusion detection technology collects and analyzes network behavior, security logs, audit data, and information on different key points in the computer system, detects abnormal behavior in time and responds, thereby protecting the security of the network system. It is an important part of the information security field. major research results. [0003] There are a lot of problems in the traditional intrusion detection system, for example, poor ability to identify unknow...

AI Technical Summary

Problems solved by technology

[0003]Traditional intrusion detection systems have a lot of problems, such as poor ability to identify unknown abnormal behaviors, high false alarm rate, occupying a lot of system resources; weak learning ability , requires a lot of manual intervention to analyze attack data; cannot defend against emerging technologies such as script attacks that are now widely used

Images