特征选择的轻量级入侵检测系统.pdf

and ComputerEngineering Applications计算机工程与应用 特征选择的轻量级入侵检测系统 田俊峰，黄红艳，常新峰 TIAN Jun-feng，HUANGHong-yan，CHANGXin—feng 河北大学数学与计算机学院，河北保定0r71002 of Science，Hebei 071002。China DepartmentMathematicsComputerUniversity，Baoding，Hebei E-mail：huanghongyan@sjzue．edu．cn TIAN selection intrusion Jun-feng，HUANGHo呜-yan，CHANGXin-feng．FeatureUghtweight detection and EngineeringApplications，2009，45(4)：111-114． redundantand I．eahlres Abstract：Featureselectioncaneliminate well．is intheenhancementofintrusion noisy advantageous a ofintrusion detection detectionanddetectionrate，thus detection basedonfeatureselectionis systems speed survey system to thefieldof conforlllSthetrendin intrusiondetection．，11lis anew intrusiondetection necessary．also paperproposes lightweisht on featureselection inthe basedfiltermode．Whetherthe datasetor ofthe of system algorithm optimization parameters thatwillraisethe classifier，areeffective detection classification thus given implementationstrategy speed，reduce interference，and theintrusiondetectionrate． improve words：feature intrusiondetection Key selection；lightweight system；filter；wrapper 摘要：特征选择能够很好地消除冗余和噪音特征，有利于提高入侵检测系统的检测速度和效果。因而对基于特征选择的入侵检 测系统进行研究是必要的，也符合入侵检测领域的发展趋势。提出了一种基于过滤器模式的轻量级入侵检测系统，无论是在数据 集的特征选择算法还是分类器的参数优化上，都给出了有效的实施策略，提高了检测速度，降低了分类干扰