AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

Evaluating Data Attribution for Text-to-Image Models

Sheng-Yu Wang, Alexei A. Efros, Jun-Yan Zhu, Richard Zhang

ICCV 2023 · Image generation › Text-to-image diffusion · approach: Personalized and region-controlled generation · 29 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
tune a linear mapping with contrastive loss on the attribution datasetDINO, CLIP, ViT, MoCo features · Object-centric and style-centric test setsRecall@10Improved
train the mapper on object and style data together rather than one (ablation)Attribution feature mapper · Object and style axesRecall@10Improved
add regularization on the affine mapping (ablation)Attribution feature mapper · All test cases averagedRecall@10Improved
use linear mapping instead of other mapping functions (ablation)Attribution feature mapper · All test cases averagedRecall@10Improved
change augmentation strength (mild vs strong) (ablation)Attribution feature mapper training · All test cases averagedRecall@10No clear effect
use copy-detection (SSCD) or style descriptor (ALADIN) features instead of general featuresPretrained feature spaces · Attribution benchmarkRecall@10Hurt
train on one distribution and test on unseen classes or sourceTuned features · ImageNet-Unseen and ArtchiveRecall@10Improved
calibrate softmax temperature and threshold for soft influence scores (ablation)Soft attribution scores · Soft influence scoresscore variance: variance < 0.0003 % at tau = 1Improved
increase fine-tuning set size to 1000 images (ablation)Exemplar-based attribution transfer to MSCOCO · MSCOCO fine-tuned modelscorrelation with benchmarkHurt
attribute media-prompted images instead of GPT-prompted imagesObject-centric attribution · Prompt typesRecall@10Hurt

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