J4 ›› 2011, Vol. 46 ›› Issue (9): 95-98.

• Articles • Previous Articles     Next Articles

Based on integrated fuzzy-neural network intrusion detection model

JIANG Jia-tao, LIU Zhi-jie*, XIE Xiao-yao   

  1. Key Laboratory of Information and Computing Science of Guizhou Province, Guizhou Normal University,
    Guiyang 550001, Guizhou, China
  • Received:2011-05-19 Online:2011-09-20 Published:2011-09-08


With increasing serious situation of network security and network defects in the agreement itself, it’s incompetent to use the traditional way of firewall. Fuzzy neural network model of integrated intrusion detection is proposed to improve the ability of intrusion prevention in network. First, the data stream is obtained from network, and the fuzzy approach is used to perform data pre-processing on characteristics of invasion. Then, the training and testing data is received by the integrated fuzzy neural network module from the data pre-processing module. Through repeated training and learning, the weights of nodes in the sub-trees converge to determine values. When training is completed, the model is used to detect the network data. The response module receives the results of fuzzy neural network module and makes the appropriate response. In the experiment, the network intrusion detection datasets, a part of KDDCUP99, are used to evaluate integrated fuzzy neural network, and compared to a single neural network model. On the whole,the result shows that fuzzy neural network ensemble method results is more stable. It was slightly reduced on false alarm rate, false negative rate and false positive rate and significantly improved on accuracy and ability of datasets generalization.

Key words:  intrusion detection; fuzzy-neural network; fuzzy integration neural network

No related articles found!
Full text



No Suggested Reading articles found!