Data Mining Based Technique for IDS Alerts Classification

TL;DR

This paper proposes a data mining technique for classifying IDS alerts to distinguish serious alerts from irrelevant ones, achieving 99.9% accuracy and reducing human intervention.

Contribution

It introduces a novel data mining-based classification method that outperforms recent techniques in IDS alert filtering.

Findings

Achieved 99.9% classification accuracy

Created a ranked alerts list based on importance

Outperformed recent data mining methods with 97% accuracy

Abstract

Intrusion detection systems (IDSs) have become a widely used measure for security systems. The main problem for those systems results is the irrelevant alerts on those results. We will propose a data mining based method for classification to distinguish serious alerts and irrelevant one with a performance of 99.9% which is better in comparison with the other recent data mining methods that have reached the performance of 97%. A ranked alerts list also created according to alerts importance to minimize human interventions.