CurvGAD : Leveraging Curvature for Enhanced Graph Anomaly Detection
Karish Grover, Geoffrey J. Gordon, Christos Faloutsos · 2025
Read CurvGAD : Leveraging Curvature for Enhanced Graph Anomaly Detection by Karish Grover, Geoffrey J. Gordon, Christos Faloutsos (2025) 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/arxiv-2502-08605.txt.
Description
Karish Grover, Geoffrey Gordon, and Christos Faloutsos used the bend of a network in 2025 to catch odd nodes that other hunters miss. Standard graph anomaly hunts rebuild a graph and flag what fits badly, but they watch only structure and traits. Bend, or curvature, marks a third kind of oddness. Their autoencoder, CurvGAD, runs two tracks: one rebuilds edge bends alone, the other rebuilds structure and traits while holding bend steady.
Splitting the tracks keeps the three odd kinds apart, so each flag can be read. On ten real data sets, mixed in kind, the method beat the best rivals by up to six and a half points. The team shares the code in public.
Catalog record
| CREATOR | Karish Grover, Geoffrey J. Gordon, Christos Faloutsos |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 81 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-08605