Neural Network based Intrusion Detection Systems

摘要: Recent Intrusion Detection Systems (IDSs) which are used to monitor real-time attacks on computer and network systems still faced with problems of low detection rate, high false positive, negative alert flooding. This paper present a Neural Network-based approach that combined supervised unsupervised learning techniques designed correct some these problems. The design is divided into two phases namely: Training Detection. In the training phase, Multiple Self–Organizing Map algorithm (SOM) was constructed capture number different input patterns, discover significant features in patterns learn how classify input. Sigmoid Activation Function (SAF) transform reasonable value (0, 1). weights were randomly assigned range (-1, +1) obtain output consistent training. SAF represented using hyperbolic tangent order increase speed make efficient. Momentum adaptive rates introduced significantly improve performance back-propagation neural network. trained lattice neuron as back propagation for monitoring intrusive activities. implemented Visual Basic.Net. An evaluation carried out Network Traffic data collected from Defence Advanced Research Projects Agency dataset consisting normal traffic. model performed by means Root Mean Square (RMS) error analysis rate 0.70, 4 layers, 8 hidden layers 2 layers. result new showed promising improved technique when compared recent best known related work.