Detection of Cyber Attacks in Networks Using Hybrid Decision Tree Technique

Abstract

Numerous advanced cyber-attacks constantly target networks because cyber security stands as the primary issue for the digital age. A new method to detect network-based cyber attacks through hybrid decision tree techniques is recommended in this study.  Conventional intrusion detection systems currently face difficulties because of the enhanced difficulty provided by modern sophisticated cyber threats. This problem demands a solution which unites the decision tree framework with ensemble learning approaches.  Our decision tree hybrid model takes advantage of tree interpretation while achieving superior results from boosting or bagging methods.  The method uses a feature selection process which helps identify significant network traffic features that lead to better detection precision and operational efficiency.  The model was tested through evaluation on a standard network traffic benchmark which established its ability to detect multiple cyber security threats.  Our hybrid decision tree model reaches remarkable performance metrics based on experimental data which reveals 96.5% accuracy along with 95.8% precision and 97.1% recall along with a 96.4% F1-score. The obtained results demonstrate that our method effectively detects cyber attacks with reliability in complicated network systems.