AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

DNA: Uncovering Universal Latent Forgery Knowledge

Jingtong Dou, Chuancheng Shi, Anqi Yi, Shiming Guo, Wenhua Wu, Yemin Wang, Li Zhang, Fei Shen, Tat-Seng Chua

ICML 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Contrastive representation theory · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use critical layers 12-18 instead of shallow layers 1-6 (ablation)DNA · SDv1.4accuracy: 62.1 -> 98.4Improved
use critical layers 12-18 instead of deep layers 19-24 (ablation)DNA · SDv1.4accuracy: 86.2 -> 98.4Improved
use critical layers 12-18 instead of transition layers 7-11 (ablation)DNA · SDv1.4accuracy: 90.7 -> 98.4Improved
use FDUs instead of excluding them (Random-Ex) (ablation)DNA · SDv1.4accuracy: 89.25 -> 98.4Improved
keep FDUs unmasked rather than masking them (ablation)DNA · base model testaccuracy: 93.1 -> 65.1Improved
keep FDUs unmasked rather than masking them (ablation)DNA · Midjourneyaccuracy: 81.6 -> 59.3Improved
mask random neurons (ablation)DNA · base model testaccuracy: 93.1 -> 91.1Hurt
mask magnitude-matched non-FDU neurons (ablation)DNA · base model testaccuracy: 93.1 -> 92.0Hurt
increase training set from 5,000 to 30,000 images (ablation)DNA · SDv1.4accuracy: 96.7 -> 95.0Hurt
apply JPEG compression quality 40DNA · perturbation robustnessAP: AP remains 92.4No clear effect
use DNA instead of MoLDGenImage Midjourney subset · GenImage Midjourney subsetaccuracy: 76.3 -> 93.7Improved
use DNA instead of MoLDHIFI-Gen · HIFI-Genaccuracy: 81.0 -> 96.4Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy