AI Paper Atlas

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

Harnessing Diffusion-Generated Synthetic Images for Fair Image Classification

Abhipsa Basu, Aviral Gupta, Abhijnya Bhat, Venkatesh Babu Radhakrishnan

AAAI 2026 · Image generation › Text-to-image diffusion · approach: Fairness and bias in generative models · 1 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 last-layer retraining on real data after vanilla SD pretraining instead of FFR full finetuningVanilla SD pipeline · Waterbirds, CelebA, UTKFaceaccuracy: outperforms FFR by an average of 2.4%Improved
cluster each group and train a Dreambooth model per clusterDreambooth · UTKFace (original)worst-group accuracy: 57.9 -> 76.0Improved
use Clustered Dreambooth instead of LoRA finetuningLoRA-finetuning · UTKFace (original)worst-group accuracy: improving upon LoRA by 7.4%Improved
cluster groups into multiple Dreambooth models (averaged over datasets and ratios)Dreambooth · three fairness benchmarks, both bias ratiosworst-group accuracy: 76.2 -> 79.1Improved
pretrain on Clustered Dreambooth images before GDRO finetuningGDRO · Waterbirds (bias ratio 0.999)worst-group accuracy: 23.5 -> 81.8Improved
pretrain on Clustered Dreambooth images before GDRO finetuningGDRO · CelebA (bias ratio 0.999)worst-group accuracy: 27.2 -> 64.4Improved
use finetuned generators (LoRA, Dreambooth variants) instead of vanilla SD for bias-conflicting UTKFace imagesClustered Dreambooth · UTKFace (bias ratio 0.999)worst-group accuracy: 67.8 -> 60.5Hurt
add CLIP-Centroid to the CLIP-Label filtering score (alpha 1 -> 0.5) (ablation)Clustered Dreambooth · UTKFace (original)worst-group accuracy: 72.7 -> 76.0Improved
use CLIP-Score filtering instead of random image selection (ablation)Clustered Dreambooth · Waterbirds (original)worst-group accuracy: 84.9 -> 88.1Improved
add the real-data finetuning stage after synthetic pretraining instead of mixing real and synthetic data in one stage (ablation)Clustered Dreambooth · CelebA (original)worst-group accuracy: 48.3 -> 84.1Improved
add SupCon loss (beta 1 -> 0.5 in both stages) (ablation)Clustered Dreambooth · UTKFace (bias ratio 0.999)worst-group accuracy: 49.6 -> 60.5Improved
add SupCon loss (beta 1 -> 0.5 in both stages) (ablation)Clustered Dreambooth · Waterbirds (original)worst-group accuracy: 88.2 -> 88.1No clear effect

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