Semise: Semi-supervised learning for severity representation in medical image
Dung T. Tran, Hung Vu, Anh Tran, Hieu Pham, Hong Nguyen, Phong Nguyen · 2025
Read Semise: Semi-supervised learning for severity representation in medical image by Dung T. Tran, Hung Vu, Anh Tran, Hieu Pham, Hong Nguyen, Phong Nguyen (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-2501-03848.txt.
Description
Dung Tran and co-authors introduce Semise, a 2025 method for learning severity aware features from medical images when labeled data is scarce. Medical image sets are often small and unevenly graded, yet downstream tools need features that reflect how severe a case looks, not just which class it falls in.
Semise blends two learning styles. Self supervised learning uses unlabeled and augmented images to teach the encoder broad structure. Supervised learning then steers those features with the labels that do exist. The mix yields richer representations than either style alone. On their tests, classification improved by 12 percent and segmentation by 3 percent over existing methods, the authors report. Better features carry into the downstream tasks clinics care about. The method targets settings where expert labels are the bottleneck, a common reality in medical imaging projects.
Catalog record
| CREATOR | Dung T. Tran, Hung Vu, Anh Tran, Hieu Pham, Hong Nguyen, Phong Nguyen |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 14 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-03848