Machine Learning-Driven Approach for DDOS Attacks Detection using Neural-Based Networks: A Proficiency Study
Abdul Ghani Ansari, Fareed Ahmed Jokhio, Muhammad Shehram Shah Syed, Hina Dharejo, Fayaz Ahmed Memon · Mehran University Research Journal of Engineering and Technology · 2025
Cyber-attacks pose significant threats to the Internet and its connected resources, causing harm to institutions and governments. Advanced technologies like cloud computing, the Internet of Things (IoT), and Artificial Intelligence (AI) have made these attacks harder to detect. Botnets, controlled by malicious actors, are at the core of many Internet attacks, including Distributed Denial of Service (DDoS) attacks. Despite years of DDoS incidents, effective defense mechanisms are lacking. In our research, we use Resilient Backpropagation, Fletcher-Powell Conjugate Gradient (FPCG), and Scaled Conjugate Gradient with three Artificial Neural Network (ANN) models: Feedforward Neural Network (FFNN), Cascade-Forward Neural Network (CFFNN), and Fitnet Neural Network (FNN) to detect DDoS attacks. The results showed that the FNN achieved higher accuracy in less time.