AI-BASED MODEL FOR CYBERSECURITY: IDENTIFYING THREATS

Abstract

The challenge of maintaining cyber-security has been intensifying due to the rapid growth in computer interconnectivity and the increasing variety of computer-based applications in recent years. To counter the rising wave of cyber threats, systems require strong and adaptive defenses. One effective solution is the use of Intrusion Detection Systems (IDS), which help identify anomalies and potential risks within computer networks. In this research, an advanced data-driven IDS is developed using Artificial Intelligence, with a particular focus on Machine Learning techniques. A new security framework, the Binary Grasshopper Optimized Twin Support Vector Machine (BGOTSVM), is introduced. This model prioritizes and ranks security features based on their importance before constructing the IDS with only the most relevant features. By reducing feature dimensions, the method not only enhances prediction accuracy for unseen data but also decreases computational cost. To evaluate its efficiency, experiments are carried out using four well-known ML methods—Decision Tree, Random Decision Forest, Random Tree, and Artificial Neural Network—then compared with existing approaches. The experimental results validate that the proposed technique can function as a reliable learning-based model for network intrusion detection and show strong potential for real-world implementation.