Anomaly Detection with Artificial Intelligence
Mohammad Nikravan, Mostafa Haghi Kashani, Sepideh Bazzaz Abkenar · 2025
The immense growth of data is inevitable in today&s;s data-driven world. These complex and sensitive data are produced by the endpoints, transmitted by the network, stored or processed in the cloud, and need robust security mechanisms. In this context, the anomaly detection method is designed to detect unusual patterns and irregularities in network traffic or stored datasets that deviate from the normal condition. These deviations could be signs of a problem, such as unexpected errors, system performance decrease, or security threats and intrusions. The anomaly detection with artificial intelligence (AI) uses machine learning (ML) and AI algorithms to identify irregularities. This anomaly detection approach does not leverage only predefined thresholds, fixed rules, and simple models but also uses complex models that continuously learn from network traffic and stored data. Therefore, AI-based anomaly detection approaches have better compatibility with dynamic environments and constantly changing patterns in detecting new irregularity patterns and are suitable for the dynamic and complex nature of modern networks. This chapter comprehensively studies AI-based anomaly detection approaches to provide network and data security. In addition, we present a classification of proposed approaches, discuss their key concepts, advantages, disadvantages, and applied tools, and highlight the main challenges related to the subject.