Lightweight Relational Embedding in Task-Interpolated Few-Shot Networks for Enhanced Gastrointestinal Disease Classification
1 st Xinliu Zhong, 2 nd Leo Hwa Liang 1, 3 rd Angela S. Koh, 4 th Yeo Si Yong 1 · 2025
Read Lightweight Relational Embedding in Task-Interpolated Few-Shot Networks for Enhanced Gastrointestinal Disease Classification by 1 st Xinliu Zhong, 2 nd Leo Hwa Liang 1, 3 rd Angela S. Koh, 4 th Yeo Si Yong 1 (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-2505-24792.txt.
Description
Xinliu Zhong, Leo Hwa Liang, Angela Koh, and Yeo Si Yong wrote a 2025 medical imaging paper on classifying gut disease from few endoscopy images. Colonoscopy is invasive, and its frames often look alike even across different conditions, which makes automated reading hard. Data are also scarce, since each exam yields limited usable images.
Their network learns from few examples, a setup built for fine grained patterns it has never seen. Task interpolation synthesizes extra training views from different instrument angles. A relational embedding module scores how images relate to class examples, and a two level attention mechanism routes the comparison. Early detection of colorectal cancer drives the work, since survival climbs when the disease is caught early. The design targets accuracy without demanding huge labeled collections.
Catalog record
| CREATOR | 1 st Xinliu Zhong, 2 nd Leo Hwa Liang 1, 3 rd Angela S. Koh, 4 th Yeo Si Yong 1 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 24 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-24792