Cybersecurity in the Era of Artificial Intelligence
Yi Qian, Rose Qingyang Hu, Shengjie Xu · 2022
Abstract The rapid and successful advances of artificial intelligence (AI) and machine learning (ML) offer security researchers and practitioners new approaches and platforms to explore and investigate challenging issues emerging in many safety-critical systems. Those AI/ML-enabled solutions have boosted the efficiency and effectiveness of multiple important security applications. While AI and ML can be adopted to identify threats more accurately and prevent cyberattacks more efficiently, cybersecurity professionals must respond to the increasingly sophisticated attacks from adversaries. Modern intelligent networking systems have been maliciously manipulated, evaded, and misled, causing significant security incidents in financial systems, cyber-physical systems, and many other critical domains. This chapter provides a background of cybersecurity in the era of AI. First, the use of AI for cybersecurity and the main workflow are introduced. They are followed by the key areas and challenges in data-driven cybersecurity, including anomaly detection, trustworthy AI, and privacy preservation. The toolbox for building secure and intelligent systems is then presented to offer the practical guide by using ML, deep learning, privacy-preserving ML, and adversarial ML. Lastly, a few data repositories for cybersecurity research, such as intrusion detection and malware detection, are introduced.