DOS intrusion attack detection by using of improved SVR
Zohreh Sadat Hosseini, Seyyed Javad Seyyed Mahdavi Chabok, Seyyed Reza Kamel · 2015
The main goal of attack detection system is classification of system activities into two basic groups of normal activities and intrusion activities. However many studies have been done in the field of intrusion detection, finding a method with min error and max accuracy is still a challenge. The main aim of this study is presenting a DOS attack intrusion detection system by using support vector regression and optimizing this algorithm by combining two algorithms of ant colony and firefly. The firefly algorithm modifies ants' positions by performing a local search and firefly performance would be optimized by decreasing random parameters. Standard KDD Cup 99 series has been used for the evaluation of intrusion detection system. This series includes 41 properties among which 9 properties have been selected. The presented system has 99.57% accuracy rate and 0.0064% error detection rate.