Abstract

In today's hyperconnected world, cybersecurity has become a top priority due to the fact that digital threats constantly evolve in terms of complexity and frequency. Cyberattacks such as malware, phishing, denial of service, and advanced persistent threats take advantage of network, system, and user behavior vulnerabilities, resulting in financial, reputational, and operational loss. This study aims to counter the growing problem of cybersecurity threats through a robust framework for detection, mitigation, and prevention of cyber attacks. Particle Swarm Optimization (PSO) is applied for feature selection to identify the most significant attributes that are shaping attack patterns. A Bi-Stacked Artificial Neural Network (Bi-Stacked ANN) is intended to learn and identify intricate relationships within the data, providing superior accuracy and stability for multi-class attack classification. The model is tested against common performance metrics to ensure its efficacy.