Astana, Kazakhstan — Researchers at Nazarbayev University have introduced a breakthrough solution in cybersecurity that could transform approaches to protecting data in the Internet of Things (IoT), cloud, and edge computing environments. Their machine learning-based anomaly detection method, published in Expert Systems with Applications, is designed to improve both the security and responsiveness of high-tech systems.
A Real-World Innovation
Modern IoT systems—including smart devices, medical instruments, and industrial sensors—are increasingly vulnerable to cyberattacks and failures. The newly developed method is designed to efficiently detect anomalies in real time, helping to prevent damage and ensure uninterrupted system operations. The researchers employed powerful machine learning algorithms such as Isolation Forest and Local Outlier Factor, which are capable of identifying anomalies like network intrusions, sensor malfunctions, or system errors.
How Does It Work?
This method processes data from IoT devices in real time to instantly detect any system irregularities. It can be applied in various domains—from monitoring smart homes to preventing breakdowns in industrial and healthcare settings. A key advantage of this technology is its ability not only to detect anomalies but also to explain their causes through Explainable AI (XAI) tools, making the process more transparent and comprehensible.