A Research On IDS Based On FSVM

In last five years,Attacks on Information System has increased 250% and induced 1000 hundred million dollars, A report has said. So network security is the emphasis in future years.In network security,Intrusion Detection System(IDS) as the important complementarity of Firewall,has much shotcomings,such as high false rate,low sensitivity in new attack measure.Now many artificial arithmetics such as NN are introduced to the intrusion detection,We hope these artificial arithmetics can solve those problem.Of course these artificial arithmetics have achieved good effect,But NN has many shortcomings which prevent it applying on the intrusion detection. We introduce Statisitical Learning Theory to it and apply Fuzzy Support Machines to the intrusion detection first.And put forward a method based Fuzzy Support Machines,We study the intrusion detection aggregate,and chose a appropriate RBF kernel fuction and get the Fuzzy Support Machines classfiers,Then we test it with DARPA has a standard intrusion data aggregate and achieved good effect.At last,I study an open source network intrusion system Bro,insert our Fsvms classfier into it,Get a artificial network intrusion system.