Neighborhood Valued Tolerance Rough Set And Its Application In Network Intrusion Detection

With the rapid development of network technology in the era of big data,while bringing convenience to human society,the network carries a large number of potential network intrusion hazards.Network intrusion detection system is an important defense tool for network information security protection,while incomplete and dynamically changing network intrusion data seriously affects the effectiveness of network intrusion detection.Effectively improving the correctness and timeliness of network intrusion detection is a necessary measure to ensure a win-win situation for both networks and netizens.Classical rough set theory has certain advantages in feature selection,knowledge discovery,and analysis of uncertain problems,but there are still limitations,including the inability to effectively handle incomplete information systems and dynamically changing data sets.This article discusses effective methods to improve the accuracy and timeliness of network intrusion detection,and improves and optimizes the classical rough set model.The new model is applied to the research of network intrusion detection,which can provide decision-making basis for effectively identifying network attacks and maintaining their own rights and interests.Firstly,aiming at the practical problems of rapid information growth and incomplete data in network intrusion detection,a feature selection algorithm based on neighborhood valued tolerance conditional entropy incremental update is proposed.Based on the granular computation of neighborhood valued tolerance,combined with the significant characteristics of conditional entropy in characterizing feature uncertainty and the degree of correlation or dependency between features,an incremental update mechanism for neighborhood valued tolerance conditional entropy is studied.Based on this update mechanism,a feature selection algorithm for incremental update of dynamic data bases is proposed.Secondly,the decision analysis method of Extreme Learning Machine based on incremental feature selection algorithm based on neighborhood valued tolerance conditional entropy is used to establish a network intrusion detection and early warning model,which can provide decision support for network intrusion detection and early warning.The newly proposed algorithm is used to reduce the dimensionality of high-dimensional network intrusion features and obtain the reduced attribute importance.Then,Extreme Learning Machine is used to evaluate the accuracy of the reduced attribute features in classifying data samples.Finally,experimental data analysis shows that the proposed algorithm can effectively improve the computational efficiency of feature selection in incomplete information systems.An application example shows that the reduced attributes have a sufficiently high classification accuracy for the sample data,while ensuring the consistency of the sample category ratio with the original dataset.This verifies the advantages of the Extreme Learning Machine decision analysis method based on the neighborhood valued tolerance conditional entropy incremental feature selection algorithm in network intrusion detection instance applications,such as computational complexity and low false alarm rate.It indicates that it can provide effective and feasible concrete methods for network information security protection.