Application of Improved Support Vector Machines in Intrusion Detection

Intrusion detection system is of most importance to network security. Support Vector Machine (SVM) is algorithm about how to solve machine learning problems under circumstance of small sample. The paper respectively applies SVM based on least square and least-square SVM improved by greedy algorithm to intrusion detection, and does simulation experiment on intrusions detection data. Experiment result shows that least-square SVM based on greedy algorithm is more suitable in intrusion detection system in circumstance that the prior knowledge is less.