Document Type : Original Article
Authors
1
Information System , Computers and Artificial Intelligence , Helwan , Cairo ,Egypt
2
Information Systems Dept., Faculty of Computers and Artificial Intelligence, Helwan University, Cairo, Egypt
3
Information System Dept., Faculty of Computers and Artificial Intelligence, Helwan University, Egypt
Abstract
Anomaly detection, which involves identifying irregular patterns that diverge from normal behavior, plays a vital role in various fields such as cybersecurity, healthcare, financial systems, and the Internet of Things (IoT). Recognizing anomalies is key to uncovering problems like fraudulent activities, system failures, or security intrusions. Traditional methods for anomaly detection, which typically rely on statistical or threshold-based techniques, are effective for low-dimensional or static data but struggle with high-dimensional, intricate, and dynamic datasets. As data complexity and volume have increased, machine learning (ML) and deep learning (DL) techniques have become essential for enhancing detection precision and adaptability. Identifying anomalies in networks is particularly important for bolstering cybersecurity, acting as a proactive approach to prevent or reduce cyber threats. With the rapid progress in Artificial Intelligence (AI), many deep learning-based approaches utilizing Autoencoders (AEs) have been created to improve network security. However, the performance of these advanced AE models varies widely, and they often lack a thorough framework for assessing critical performance metrics that impact detection accuracy.
Keywords
Main Subjects
)
مشاهدة على SCiNiTO