Evaluation effectiveness of intrusion detection system with reduced dimension using data mining classification tools

Intrusions detections systems from point of view of security policy are a second line of defense; they have a supervisory role to observe the activities of our network or hosts to identify attacks in real time. In our days, electronics attacks can cause a very destructive damage for nations which make necessary the use of completed security policy to minimize the potential threats. IDS it is a very important element to resist against this vulnerability, in our works, we use a wired data base Knowledge Discovery Data Mining (KDD) CUP 99 and a Data Mining Tools Waikato Environment for Knowledge Analysis (WEKA) to combine the advantages of an intrusion detection algorithm (PART) and two techniques of Dimensionality Reduction(best first search and genetic search), to evaluate our works, we applied the proposed combined technique, and we check the results by using a several evaluations parameters. The results show that a very high detection rate for certain attacks types and highest sensitivities with the hybrid dimensionality reduction method.