ANIMATE: Unsupervised Attributed Graph Anomaly Detection with Masked Graph Transformers
Jingtao Hu, Yi Zhang, Chengzhang Zhu, Changsheng Hou · 2026
Read ANIMATE: Unsupervised Attributed Graph Anomaly Detection with Masked Graph Transformers by Jingtao Hu, Yi Zhang, Chengzhang Zhu, Changsheng Hou (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13211153.txt.
Description
Jingtao Hu, Yi Zhang, Chengzhang Zhu, and Changsheng Hou present a method to spot odd nodes in labeled networks without any training labels, in a 2026 paper. Sensor networks and the internet of things throw up graphs where each node carries traits; the rare odd ones matter most. Older graph methods look only at each node's neighborhood and blur differences as they stack layers. Their system, ANIMATE, swaps in transformer blocks that see the whole graph at once.
A masking step hides node traits and makes the model rebuild them, teaching it the normal pattern. A self-paced scheme then sharpens its sense of the rare. On real benchmark graphs with genuine odd nodes, the system beat the leading rivals. The authors target reliability work for sensor fleets and connected devices, where catching one bad node early can save the network.
Catalog record
| CREATOR | Jingtao Hu, Yi Zhang, Chengzhang Zhu, Changsheng Hou |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |