LGDC: Latent Graph Diffusion via Spectrum-Preserving Coarsening
Nagham OsmanUniversity College LondonLondon, UK [email protected], Keyue JiangUniversity College LondonLondon, UKDavide BuffelliMediaTek ResearchLondon, UKXiaowen DongUniversity of OxfordOxford, UKLaura ToniUniversity College LondonLondon, UK · 2025
Read LGDC: Latent Graph Diffusion via Spectrum-Preserving Coarsening by Nagham OsmanUniversity College LondonLondon, UK [email protected], Keyue JiangUniversity College LondonLondon, UKDavide BuffelliMediaTek ResearchLondon, UKXiaowen DongUniversity of OxfordOxford, UKLaura ToniUniversity College LondonLondon, UK (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-2512-01190.txt.
Description
Nagham Osman and four coauthors present LGDC in a 2025 paper on graph generation. Generating realistic graphs matters across science, from molecules to social networks. Existing methods come in two broad kinds. Autoregressive models grow a graph step by step, while one-shot models, such as diffusion, produce the whole graph at once.
LGDC performs latent graph diffusion with spectrum-preserving coarsening. Coarsening shrinks a graph while keeping its spectrum, a summary of its structure, intact. Working in a latent space and coarsening carefully could make diffusion faster without losing the traits that make a graph realistic.
Catalog record
| CREATOR | Nagham OsmanUniversity College LondonLondon, UK [email protected], Keyue JiangUniversity College LondonLondon, UKDavide BuffelliMediaTek ResearchLondon, UKXiaowen DongUniversity of OxfordOxford, UKLaura ToniUniversity College LondonLondon, UK |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-01190