Adaptive Forensic Feature Refinement via Intrinsic Importance Perception DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; November 10-14, 2026; Rio de Janeiro, Brazil
Jiazhen Yang, Junjun Zheng, Kejia Chen, Xiangheng Kong, Jie Lei, Zunlei Feng, Bingde Hu, Yang Gao · 2026
Read Adaptive Forensic Feature Refinement via Intrinsic Importance Perception DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; November 10-14, 2026; Rio de Janeiro, Brazil by Jiazhen Yang, Junjun Zheng, Kejia Chen, Xiangheng Kong, Jie Lei, Zunlei Feng, Bingde Hu, Yang Gao (2026) 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-2604-16879.txt.
Description
Yang and seven colleagues tackle synthetic-image detection, where tools trained on known fakes often fail on unfamiliar generators. Their I2P framework picks out the vision-model layer carrying the strongest forgery clues. It then limits task tuning to a low-sensitivity part of the model so pretrained knowledge stays intact, giving steadier results on fresh sources.
Catalog record
| CREATOR | Jiazhen Yang, Junjun Zheng, Kejia Chen, Xiangheng Kong, Jie Lei, Zunlei Feng, Bingde Hu, Yang Gao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 43848 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.16879