期刊名称：International Journal of Electrical and Computer Engineering
电子版ISSN：2088-8708
出版年度：2013
卷号：3
期号：5
页码：708-712
语种：English
出版社：Institute of Advanced Engineering and Science (IAES)
摘要：With increasing tendency of users to distributed computer systems in comparison with concentrat-ed systems, intrusion into such systems has emerged as a serious challenge. Since techniques of intrusion into systems are being intelligent, it seems necessary to use intelligent methods to en-counter them. Success of the intrusion systems depends on the strategy employed in these sys-tems for attack detection. Application of eXtended Classifier Systems (XCS) is proposed in this paper for detection of intrusions to databases. The extended classifier systems which are known as one of the most successful types of learning agents create a set of stochastic rules and com-plete them based on the methods inspired from human learning process. Thereby, they can grad-ually get a comprehensive understanding of the environment under study which enables them to predict the correct answer at an acceptable accuracy once encountered with new issues. The method suggested in this paper an improved version of extended classifier systems is “trained” using a set of existing examples in order to identify and avoid attempts to intrude computer sys-tems during phases of application and encountering these attempts. The proposed method has been tested on several problems to demonstrate its performance while its results indicate a 91% detection of various known intrusions to the databases. DOI: http://dx.doi.org/10.11591/ijece.v3i5.4034
其他摘要：With increasing tendency of users to distributed computer systems in comparison with concentrat-ed systems, intrusion into such systems has emerged as a serious challenge. Since techniques of intrusion into systems are being intelligent, it seems necessary to use intelligent methods to en-counter them. Success of the intrusion systems depends on the strategy employed in these sys-tems for attack detection. Application of eXtended Classifier Systems (XCS) is proposed in this paper for detection of intrusions to databases. The extended classifier systems which are known as one of the most successful types of learning agents create a set of stochastic rules and com-plete them based on the methods inspired from human learning process. Thereby, they can grad-ually get a comprehensive understanding of the environment under study which enables them to predict the correct answer at an acceptable accuracy once encountered with new issues. The method suggested in this paper an improved version of extended classifier systems is “trained” using a set of existing examples in order to identify and avoid attempts to intrude computer sys-tems during phases of application and encountering these attempts. The proposed method has been tested on several problems to demonstrate its performance while its results indicate a 91% detection of various known intrusions to the databases. DOI: http://dx.doi.org/10.11591/ijece.v3i5.4034
Loading...
