AI-Based Network Intrusion Detection with Hyperparameter Optimization on the Realistic CSE-CICIDS2018 Dataset Using Cloud Computing

Artificial intelligence (AI) is one of the most prominent modern technologies that has revolutionized the world of technology. It enables machines to mimic human behavior and learn from the data around them, enabling them to make intelligent decisions and interact in ways that closely resemble human behavior. One of the most significant areas that has benefited from this tremendous development is information security, particularly in combating cyberattacks and detecting suspicious activities that could threaten computer systems and networks. Intrusion detection systems (IDS) are essential tools used in this context, as they monitor and analyze network traffic to identify any unauthorized attempts to access or manipulate data. With the ever-increasing sophistication of cyberattacks, the need to develop smarter and more adaptive methods has emerged. Hence, relying on AI has become an ideal solution for enhancing the performance of IDSs. AI doesn't just analyze data; it continually learns from it and improves its capabilities in detecting new threats that may not have been previously known. Among the AI techniques that have proven highly effective in this field are artificial neural networks (ANNs). These are designed to mimic the way the human brain analyzes information and makes decisions. They are capable of processing massive amounts of complex data and detecting hidden patterns that may indicate a cyberattack. A proposed intelligent system based on AI, specifically ANNs, was built to detect botnet attacks, a type of cyberattack considered one of the most dangerous threats, especially to the financial and banking sectors. A group of compromised devices operate in a coordinated and organized manner to launch cyberattacks that cause significant damage to targeted systems. The proposed system was trained using a recent, realistic security dataset known as CSE-CIC-IDS2018, which was created by the Canadian Institute of Cybersecurity in 2018 on the Amazon cloud computing platform. This dataset represents real-life scenarios of multiple cyberattacks. This system demonstrated high efficiency in detecting botnet attacks, achieving highly accurate results in classifying normal and malicious activities with near-perfect accuracy, along with a strong ability to distinguish between actual threats and false alarms, making it a highly reliable system. The system is also characterized by its flexibility and the ability to be implemented on an unlimited number of devices, making it suitable for use in various environments, whether traditional networks or cyber systems integrated into physical environments such as factories and power grids, or even in environments that require real-time monitoring and analysis of network traffic. Based on the above, it can be said that the proposed system represents a qualitative leap in the field of information security, combining the latest artificial intelligence technologies with real-world data, providing an integrated solution that is accurate, fast, and reliable in detecting complex cyber threats. This enhances organizations' ability to protect their data and infrastructure from the growing dangers of the online world.

کلیدواژه ها:

Artificial Intelligence ، CSE-CICIDS2018 Dataset ، Cybersecurity ، Hyper-Parameter Optimization ، Network Intrusion Detection ،