aiqbench analysis
This paper challenges the trend toward increasingly complex architectures in streaming graph anomaly detection by introducing SimpleCount, a lightweight baseline requiring no parameter fitting. The study demonstrates that sophisticated models like SLADE often offer negligible performance gains while consuming 23 to 133 times more compute time than this simple reference approach. Researchers should adopt such strong one-feature baselines and report their results alongside these computational costs before claiming architectural improvements.
Beat the Counter First: A Baseline for Temporal-Graph Anomaly Detectors
arXiv:2608.15965v1 Announce Type: new Abstract: Progress in streaming, edge-level graph anomaly detection (GAD) has been marked by increasingly elaborate architectures, from count-min-sketch chi square tests to memory-augmented attention networks. Yet the empirical gains attributable to this added complexity have not been systematically evaluated. We propose SimpleCount, a reference with no parameter fitting that selects one scalar feature per dataset from a fixed pool of counts, recencies, first-occurrence indicators, and count-derived transforms. We compare SimpleCount with two temporal-graph detector models and an IsoForest control fitted to the complete feature vector across five public datasets and one synthetic dataset. SimpleCount matches or exceeds SLADE on three of six datasets and exceeds IsoForest on all six. We report paired statistical tests and five-seed SLADE evaluations. SLADE requires 23 to 133x more wall-clock time than SimpleCount. On Synth-Triangle and an additional Synth-Quad probe, pre-event structural scores recover the planted signal at AUC up to 0.955, while all evaluated detector models remain near random. The benefit of complexity is dataset-dependent, and every claimed gain should be reported against a strong one-feature reference together with its compute cost.
Abstract:Progress in streaming, edge-level graph anomaly detection (GAD) has been marked by increasingly elaborate architectures, from count-min-sketch chi square tests to memory-augmented attention networks. Yet the empirical gains attributable to this added complexity have not been systematically evaluated. We propose SimpleCount, a reference with no parameter fitting that selects one scalar feature per dataset from a fixed pool of counts, recencies, first-occurrence indicators, and count-derived transforms. We compare SimpleCount with two temporal-graph detector models and an IsoForest control fitted to the complete feature vector across five public datasets and one synthetic dataset. SimpleCount matches or exceeds SLADE on three of six datasets and exceeds IsoForest on all six. We report paired statistical tests and five-seed SLADE evaluations. SLADE requires 23 to 133x more wall-clock time than SimpleCount. On Synth-Triangle and an additional Synth-Quad probe, pre-event structural scores recover the planted signal at AUC up to 0.955, while all evaluated detector models remain near random. The benefit of complexity is dataset-dependent, and every claimed gain should be reported against a strong one-feature reference together with its compute cost.
Machine Learning (cs.LG)
arXiv:2608.15965 [cs.LG]
(or arXiv:2608.15965v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2608.15965 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 271 words
Related coverage
SiST-GNN: Simultaneous Spatial-Temporal Message Passing for Dynamic Graph Representation Learning
arXiv · Sep 22, 2026
The Selection Rule Decides the Winner: A Pre-Registered Audit of Open-Set Graph Anomaly Detection
arXiv · Sep 30, 2026
Signal2Symbol: Neuro-Symbolic Temporal Reasoning for Explainable Physiological Time-Series Anomaly Detection
arXiv · Sep 24, 2026
TH-GNN: Heterogeneous Temporal Graph Neural Networks for LLM-Agent Shilling Attack Detection
arXiv · Aug 24, 2026