Volume 17, Number 1
FSHAHA: Feature Selection using Hybrid Ant Harris Algorithm for IoT Network Security Enhancement
Authors
Priyanka and Anoop Kumar, Banasthali Vidyapith, India
Abstract
Enhancing machine learning model performance involves selecting relevant features, particularly in high-dimensional datasets. This paper proposes a hybrid method named the Multi-Objective Ant Chase algorithm, which integrates Ant Colony Optimization (ACO) and Harris Hawk Optimization (HHO) for effective feature selection. ACO excels at exploring large search spaces using pheromone-guided navigation, while HHO focuses on targeted search with adaptive hunting tactics. Conventional algorithms, such as Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Grey Wolf Optimizer (GWO), and Monarch Butterfly Optimization (MBO), often face premature convergence in high-dimensional, sparse datasets, becoming stuck in local optima. Unlike these, the ACO and HHO combination balances exploration and exploitation efficiently. ACO’s broad search capability complements HHO’s fast convergence, providing robust global optimization. Experimental results indicate that the Multi-Objective Ant Chase algorithm outperforms individual ACO, HHO, and other comparative algorithms across metrics like Accuracy, Sensitivity, Specificity, False Alarm Rate, and Detection Rate.
Keywords
Internet of things, Feature Selection, Ant-colony algorithm, Harris hawk algorithm, Network security