From Pixels to Insights: Image Datasets for AI/ML in Software-Defined Networking
Pranav Pant, Rishikant Mallick, Santosh Kumar Sahu, Lalit Kumar Vashishtha · 2023
In the modern landscape of pervasive digital data, the need for robust intrusion detection systems (IDS) has become paramount to safeguard networks from malicious activities. This research explores the synergistic integration of big data analytics and AI/ML techniques, with a focus on leveraging Apache Spark as a platform for intrusion detection. The study delves into the core principles underlying intelligent IDS systems enhanced by AI, leading to heightened accuracy and adaptability in countering dynamic cyber threats. The research introduces an innovative SDN DDoS attack image dataset, refined to optimize machine learning and deep learning models for IDS. Through novel image generation methodologies, the dataset is enriched, substantially enhancing model performance. Comparative analyses are conducted, contrasting the original dataset with the refined versions, utilizing diverse learning strategies. The findings underscore the substantial advancements achieved, revealing notably improved model accuracy. The research also hints at the potential adoption of convolutional neural networks to further elevate classification capabilities, instigating prospective investigations into optimizing IDS models. This study not only contributes to the evolution of intrusion detection systems but also sheds light on the symbiotic interplay between big data analytics, AI/ML, and the realm of cybersecurity.