HTG-GCL: Leveraging Hierarchical Topological Granularity from Cellular Complexes for Graph Contrastive Learning
Qirui Ji, Bin Qin, Yifan Jin, Yunze Zhao, Chuxiong Sun, Changwen Zheng, Jianwen Cao, Jiangmeng Li · 2025
Read HTG-GCL: Leveraging Hierarchical Topological Granularity from Cellular Complexes for Graph Contrastive Learning by Qirui Ji, Bin Qin, Yifan Jin, Yunze Zhao, Chuxiong Sun, Changwen Zheng, Jianwen Cao, Jiangmeng Li (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-2512-02073.txt.
Description
Graph contrastive learning teaches a model what makes a graph itself by comparing altered views of it that should still match. Qirui Ji and seven co-authors present HTG-GCL in 2025, a method that reads a graph's topology at several levels of detail, from fine to coarse.
Existing methods that alter graph structure often miss the topological patterns a task actually needs, and different tasks need different levels of detail. The authors extract hierarchical topological granularity from cellular complexes, richer structures built on top of graphs, and feed it into the contrast. Respecting topology at the right scale should yield graph representations that transfer better across chemistry, biology, and network tasks.
Catalog record
| CREATOR | Qirui Ji, Bin Qin, Yifan Jin, Yunze Zhao, Chuxiong Sun, Changwen Zheng, Jianwen Cao, Jiangmeng Li |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 38 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-02073