Intrusion Detection Utilizing an Ant Colony Optimization-Based Feature Selection and the XGBoost Classifier

Shweta Bhardwaj, Seema Rawat, Hima Bindu Maringanti

2026

Data confidence 70 medium

How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

The Internet of Things (IoT) continues to expand dramatically, connecting a growing number of smart devices such as home automation systems and wearables. However, this growth also introduces significant cybersecurity risks, as attackers increasingly exploit vulnerabilities in these interconnected devices. Protecting IoT networks requires comprehensive Intrusion Detection Systems (IDSs) that can intelligently identify and mitigate malicious activities. The proposed approach integrates dimensionality reduction through Principal Component Analysis (PCA) to streamline data, feature selection using Ant Colony Optimization (ACO) to identify relevant indicators, and classification through the Extreme Gradient Boosting (XGBoost) algorithm for accurate threat detection. The proposed approach achieved far superior results compared to existing IDS methods on three different datasets: 99.2% accuracy, 99.6% precision, 98.8% recall, and 99.2% F1- score on NSL-KDD, 99.3% accuracy, 92.8% precision, 99% recall and 95.8% F1-score on UNSW- NB15, and 99.9% accuracy, 99.5% precision, 99.8% recall, and 99.7% F1-score on CIC-IDS.