Enhanced network attack detection using SK-DenseNet and optimized artificial fish swarm algorithm
D Jayanarayana Reddy, Sakthi Govindaraju, R Mahaveerakannan, Achyut Shankar, Prasanalakshmi Balaji · Journal of Cyber Security Technology · 2025
The role of artificial intelligence in enhancing cybersecurity and threat detection, including the ability to identify and stop assaults, including adversarial attacks, zero-day vulnerabilities, and network breaches, is promising. Studies across various disciplines, including driverless cars, the Internet of Things, 5 G networks, and Industry 5.0, showcase artificial intelligence’s adaptation to handle many security challenges. Modern techniques such as federated learning, blockchain integration, and transformer-based models enable improved, more accurate threat detection systems in real time. The administration of enormous volumes of data, real-time processing ability, and guarantees of privacy and security are challenging. This work investigates network attack detection leveraging the SK-DenseNet model. Based on deep learning, hyperparameter tuning with an artificial fish swarm optimization method. The experimental evaluations reveal that the detection accuracy, recall, and false-positive reduction are much higher when SK-DenseNet is coupled with the optimal artificial fish swarm technique. Our system discovers overlapping and relevant factors influencing many artificial intelligence models, clarifying comparable trends across several detection techniques. By giving security analysts multiple explanations, our paradigm enables them to make better threat identification and mitigation decisions.