Salvato in:
| Autori principali: | , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | https://arxiv.org/abs/2510.12094 |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866915553160986624 |
|---|---|
| author | Zhang, Heng Zhang, Tianyi Liu, Zijun Shi, Yuling Shen, Yaomin You, Haochen Hu, Haichuan Gan, Lubin Huang, Jin |
| author_facet | Zhang, Heng Zhang, Tianyi Liu, Zijun Shi, Yuling Shen, Yaomin You, Haochen Hu, Haichuan Gan, Lubin Huang, Jin |
| contents | Text-attributed graphs are widely used across domains, offering rich opportunities for zero-shot learning via graph-text alignment. However, existing methods struggle with tasks requiring fine-grained pattern recognition, particularly on heterophilic graphs. Through empirical and theoretical analysis, we identify an \textbf{over-abstraction problem}: current approaches operate at excessively large hyperbolic radii, compressing multi-scale structural information into uniform high-level abstractions. This abstraction-induced information loss obscures critical local patterns essential for accurate predictions. By analyzing embeddings in hyperbolic space, we demonstrate that optimal graph learning requires \textbf{faithful preservation} of fine-grained structural details, better retained by representations positioned closer to the origin. To address this, we propose \textbf{H4G}, a framework that systematically reduces embedding radii using learnable block-diagonal scaling matrices and Möbius matrix multiplication. This approach restores access to fine-grained patterns while maintaining global receptive ability with minimal computational overhead. Experiments show H4G achieves state-of-the-art zero-shot performance with \textbf{12.8\%} improvement on heterophilic graphs and \textbf{8.4\%} on homophilic graphs, confirming that radius reduction enables faithful multi-scale representation for advancing zero-shot graph learning. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_12094 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | H4G: Unlocking Faithful Inference for Zero-Shot Graph Learning in Hyperbolic Space Zhang, Heng Zhang, Tianyi Liu, Zijun Shi, Yuling Shen, Yaomin You, Haochen Hu, Haichuan Gan, Lubin Huang, Jin Machine Learning Graphics Text-attributed graphs are widely used across domains, offering rich opportunities for zero-shot learning via graph-text alignment. However, existing methods struggle with tasks requiring fine-grained pattern recognition, particularly on heterophilic graphs. Through empirical and theoretical analysis, we identify an \textbf{over-abstraction problem}: current approaches operate at excessively large hyperbolic radii, compressing multi-scale structural information into uniform high-level abstractions. This abstraction-induced information loss obscures critical local patterns essential for accurate predictions. By analyzing embeddings in hyperbolic space, we demonstrate that optimal graph learning requires \textbf{faithful preservation} of fine-grained structural details, better retained by representations positioned closer to the origin. To address this, we propose \textbf{H4G}, a framework that systematically reduces embedding radii using learnable block-diagonal scaling matrices and Möbius matrix multiplication. This approach restores access to fine-grained patterns while maintaining global receptive ability with minimal computational overhead. Experiments show H4G achieves state-of-the-art zero-shot performance with \textbf{12.8\%} improvement on heterophilic graphs and \textbf{8.4\%} on homophilic graphs, confirming that radius reduction enables faithful multi-scale representation for advancing zero-shot graph learning. |
| title | H4G: Unlocking Faithful Inference for Zero-Shot Graph Learning in Hyperbolic Space |
| topic | Machine Learning Graphics |
| url | https://arxiv.org/abs/2510.12094 |