Training-Free Dual Hyperbolic Adapters for Better Cross-Modal Reasoning
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912755755253760 |
|---|---|
| author | Zhang, Yi Cheng, Chun-Wun He, Junyi Yu, Ke Tang, Yushun Schönlieb, Carola-Bibiane He, Zhihai Aviles-Rivero, Angelica I. |
| author_facet | Zhang, Yi Cheng, Chun-Wun He, Junyi Yu, Ke Tang, Yushun Schönlieb, Carola-Bibiane He, Zhihai Aviles-Rivero, Angelica I. |
| contents | Recent research in Vision-Language Models (VLMs) has significantly advanced our capabilities in cross-modal reasoning. However, existing methods suffer from performance degradation with domain changes or require substantial computational resources for fine-tuning in new domains. To address this issue, we develop a new adaptation method for large vision-language models, called \textit{Training-free Dual Hyperbolic Adapters} (T-DHA). We characterize the vision-language relationship between semantic concepts, which typically has a hierarchical tree structure, in the hyperbolic space instead of the traditional Euclidean space. Hyperbolic spaces exhibit exponential volume growth with radius, unlike the polynomial growth in Euclidean space. We find that this unique property is particularly effective for embedding hierarchical data structures using the Poincaré ball model, achieving significantly improved representation and discrimination power. Coupled with negative learning, it provides more accurate and robust classifications with fewer feature dimensions. Our extensive experimental results on various datasets demonstrate that the T-DHA method significantly outperforms existing state-of-the-art methods in few-shot image recognition and domain generalization tasks. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2512_08820 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Training-Free Dual Hyperbolic Adapters for Better Cross-Modal Reasoning Zhang, Yi Cheng, Chun-Wun He, Junyi Yu, Ke Tang, Yushun Schönlieb, Carola-Bibiane He, Zhihai Aviles-Rivero, Angelica I. Computer Vision and Pattern Recognition Artificial Intelligence Recent research in Vision-Language Models (VLMs) has significantly advanced our capabilities in cross-modal reasoning. However, existing methods suffer from performance degradation with domain changes or require substantial computational resources for fine-tuning in new domains. To address this issue, we develop a new adaptation method for large vision-language models, called \textit{Training-free Dual Hyperbolic Adapters} (T-DHA). We characterize the vision-language relationship between semantic concepts, which typically has a hierarchical tree structure, in the hyperbolic space instead of the traditional Euclidean space. Hyperbolic spaces exhibit exponential volume growth with radius, unlike the polynomial growth in Euclidean space. We find that this unique property is particularly effective for embedding hierarchical data structures using the Poincaré ball model, achieving significantly improved representation and discrimination power. Coupled with negative learning, it provides more accurate and robust classifications with fewer feature dimensions. Our extensive experimental results on various datasets demonstrate that the T-DHA method significantly outperforms existing state-of-the-art methods in few-shot image recognition and domain generalization tasks. |
| title | Training-Free Dual Hyperbolic Adapters for Better Cross-Modal Reasoning |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence |
| url | https://arxiv.org/abs/2512.08820 |