Nasim Ferdosian

Laura Jiang, Reza Ryan, Qian Li et al.

Single-turn safety evaluation is a poor proxy for real fraud defense, where attackers escalate across multiple rounds. This paper evaluates fraud defenders under replay and adaptive multi-round attacks and measures when a defender refuses, not just whether it eventually refuses. On a frozen multi-round suite built from Fraud-R1, graph-context defenders improve early safe refusal relative to text-only baselines under both replay and adaptive fraud pressure, but they also produce substantially more benign over-refusal. Direct probing of the trained graph encoder, together with paired shuffle-risk ablations on both fraud and benign sides replicated across two seeds on the Qwen-1.5B backbone, localises this cost to how the defender LLM consumes structured context rather than to graph-encoder quality: the encoder cleanly separates fraud from benign, while the LLM responds primarily to the presence of structured graph fields and only secondarily, and asymmetrically, to risk-score magnitude. Temporal graph context is directionally stronger than static and significantly better grounded, but is not yet conclusively superior on the main refusal metrics. The contribution is evaluative and measurement-oriented: robust fraud assessment must be multi-round, must report refusal timing, must account for benign false positives alongside fraud-side safety gains, and must localize observed costs to the graph signal or to how the LLM consumes it.

Laura Jiang, Reza Ryan, Qian Li et al.

Anomaly detection is a critical task in cybersecurity, where identifying insider threats, access violations, and coordinated attacks is essential for ensuring system resilience. Graph-based approaches have become increasingly important for modeling entity interactions, yet most rely on homogeneous and static structures, which limits their ability to capture the heterogeneity and temporal evolution of real-world environments. Heterogeneous Graph Neural Networks (HGNNs) have emerged as a promising paradigm for anomaly detection by incorporating type-aware transformations and relation-sensitive aggregation, enabling more expressive modeling of complex cyber data. However, current research on HGNN-based anomaly detection remains fragmented, with diverse modeling strategies, limited comparative evaluation, and an absence of standardized benchmarks. To address this gap, we provide a comprehensive survey of HGNN-based anomaly detection methods in cybersecurity. We introduce a taxonomy that classifies approaches by anomaly type and graph dynamics, analyze representative models, and map them to key cybersecurity applications. We also review commonly used benchmark datasets and evaluation metrics, highlighting their strengths and limitations. Finally, we identify key open challenges related to modeling, data, and deployment, and outline promising directions for future research. This survey aims to establish a structured foundation for advancing HGNN-based anomaly detection toward scalable, interpretable, and practically deployable solutions.

Xenofon Vasilakos, Navid Nikaein, Dean H Lorenz et al.

Fifth Generation (5G) networks are envisioned to be fully autonomous in accordance to the ETSI-defined Zero touch network and Service Management (ZSM) concept. To this end, purpose-specific Machine Learning (ML) models can be used to manage and control physical as well as virtual network resources in a way that is fully compliant to slice Service Level Agreements (SLAs), while also boosting the revenue of the underlying physical network operator(s). This is because specially designed and trained ML models can be both proactive and very effective against slice management issues that can induce significant SLA penalties or runtime costs. However, reaching that point is very challenging. 5G networks will be highly dynamic and complex, offering a large scale of heterogeneous, sophisticated and resource-demanding 5G services as network slices. This raises a need for a well-defined, generic and step-wise roadmap to designing, building and deploying efficient ML models as collaborative components of what can be defined as Cognitive Network and Slice Management (CNSM) 5G systems. To address this need, we take a use case-driven approach to design and present a novel Integrated Methodology for CNSM in virtualized 5G networks based on a concrete eHealth use case, and elaborate on it to derive a generic approach for 5G slice management use cases. The three fundamental components that comprise our proposed methodology include (i) a 5G Cognitive Workflow model that conditions everything from the design up to the final deployment of ML models; (ii) a Four-stage approach to Cognitive Slice Management with an emphasis on anomaly detection; and (iii) a Proactive Control Scheme for the collaboration of different ML models targeting different slice life-cycle management problems.