The Internet of Things (IoT) has been increasingly adopted in domains such as smart infrastructure, healthcare, supply chain, transportation, and many others. However, the constrained computational resources of these devices make conventional security approaches against security threats not applicable. This limitation emphasizes the need to explore new approaches, specifically tailored to these kinds of devices. To increase protection against cyberattacks in IoT devices, Intrusion Detection Systems (IDSs) are considered an effective approach. Machine Learning (ML) techniques can be combined with Federated Learning to enhance the privacy and scalability of these systems. Many IDSs have been introduced, but there is a research gap concerning Host Intrusion Detection System (HIDS), which is the primary focus of our current work. Additionally, existing research predominantly focuses on the application of ML techniques and their evaluation, with limited attention to real-world implementation. We propose a lightweight HIDS that relies on the analysis of system call traces to detect malicious activities. The proposed HIDS achieved an accuracy rate of approximately 98%. Finally, by using eXplainable Artificial Intelligence methods, we sought to provide explanations for these these results.
Machine learning is increasingly used in cybersecurity, for example, for anomaly and intrusion detection in microservice and Internet of Things environments. Guidance on securing the machine learning operations (MLOps) lifecycle, termed MLSecOps or SecMLOps, recommends continuous monitoring, a versioned model registry, and lifecycle integration of training, evaluation, and serving. This layer is typically built per deployment, as model-specific code rather than a shared framework. We present AICybOps, a framework that systematizes these practices so the operational features and security properties a deployed model needs are provided by design. It wraps trained models behind a uniform lifecycle interface, exposes a monitor control application programming interface (API) for long-running prediction loops, re-loads models by registered name and version across hosts, and runs durable asynchronous training and evaluation, so a developer implements only the model and inherits the rest. These four design features constitute the framework’s current version; an inheritance-and-expansion criterion governs its contents and keeps it open to further security operations. Because the framework addresses a deployed model by registered version, tracks its runs automatically, and holds the registry behind the service boundary, it yields five security properties: lineage, reproducible replay, an audit trail, private provenance, and integrity of model identity.