Research on SVDD Network Intrusion Detection of the Optimal Feature Selection for Particle Swarm
Focusing on the problem about the higher dimensionality of sample set in the intrusion detection, propose an optimized method of support vector data description (SVDD) based on particle swarm optimization (PSO) and apply it to the intrusion detection of network exception. This method adopts PSO to eliminate the superfluous parameters in SVDD and carries out dimension reduction to data; then, establish the super sphere model to detect the network intrusion data and output the results of intrusion detection. Carry out the simulation experiment based on the standard detection data set of KDD CUP' 99, and the result shows that this method, comparing with the traditional SVDD, can effectively improve the detection ratio with a smaller amount of calculation.