Pith. sign in

REVIEW 7 cited by

Finetuning Text-to-Image Diffusion Models for Fairness

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2311.07604 v2 pith:JLRIXX7U submitted 2023-11-11 cs.LG cs.AIcs.CVcs.CY

classification cs.LGcs.AIcs.CVcs.CY
keywords diffusionfinetuningadjustedbiasesfairnessgendermethodalignment
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
abstract

The rapid adoption of text-to-image diffusion models in society underscores an urgent need to address their biases. Without interventions, these biases could propagate a skewed worldview and restrict opportunities for minority groups. In this work, we frame fairness as a distributional alignment problem. Our solution consists of two main technical contributions: (1) a distributional alignment loss that steers specific characteristics of the generated images towards a user-defined target distribution, and (2) adjusted direct finetuning of diffusion model's sampling process (adjusted DFT), which leverages an adjusted gradient to directly optimize losses defined on the generated images. Empirically, our method markedly reduces gender, racial, and their intersectional biases for occupational prompts. Gender bias is significantly reduced even when finetuning just five soft tokens. Crucially, our method supports diverse perspectives of fairness beyond absolute equality, which is demonstrated by controlling age to a $75\%$ young and $25\%$ old distribution while simultaneously debiasing gender and race. Finally, our method is scalable: it can debias multiple concepts at once by simply including these prompts in the finetuning data. We share code and various fair diffusion model adaptors at https://sail-sg.github.io/finetune-fair-diffusion/.

Discussion (0). Sign in to comment.

Forward citations

Cited by 7 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. FairFlow: Demystifying and Mitigating Stereotype Bias in Text-to-Image Diffusion Transformers

    cs.CV 2026-07 conditional novelty 7.0 of 10

    Bias in MM-DiTs is mediated by sparse stage-wise semantic binding hubs, and sparse inference-time steering at those hubs mitigates gender, race, and intersectional stereotypes with low overhead.

  2. FAROS: Fair Graph Generation via Attribute Switching Mechanisms

    cs.LG 2025-07 conditional novelty 6.0 of 10

    FAROS switches sensitive attributes on an optimal fraction of nodes at an optimal diffusion step to produce fairer generated graphs with little accuracy loss.

  3. Kernel-based Unsupervised Embedding Alignment for Enhanced Visual Representation in Vision-language Models

    cs.CV 2025-06 conditional novelty 6.0 of 10

    Fine-tuning CLIP's visual encoder to match DINOv2's kernel-based similarity structure improves its fine-grained visual perception while preserving its alignment to text.

  4. Multi-Group Proportional Representation for Text-to-Image Models

    cs.CV 2025-05 conditional novelty 6.0 of 10

    The authors apply the MPR metric (an integral probability metric) to text-to-image generation, derive tractable forms for linear and decision-tree function classes, and use it as a fine-tuning objective that reduces i...

  5. BioPro: Towards Difference-Aware Gender Fairness for Vision-Language Models

    cs.AI 2025-11 conditional novelty 5.0 of 10

    BioPro uses orthogonal projection on a gender-variation subspace to selectively debias vision-language models, reducing gender bias in neutral contexts while preserving explicit gender cues.

  6. FairJudge: Abstention-Aware Multimodal Judges for Fairness and Alignment Evaluation in Text-to-Image Models

    cs.CV 2025-10 conditional novelty 5.0 of 10

    An abstention-aware, label-constrained MLLM protocol plus two synthetic benchmarks improves social-attribute and alignment evaluation of text-to-image models on most tested attributes over CLIP/DeepFace.

  7. Can we Debias Social Stereotypes in AI-Generated Images? Examining Text-to-Image Outputs and User Perceptions

    cs.HC 2025-05 conditional novelty 5.0 of 10

    A rubric-based Social Stereotype Index shows prompt refinement lowers measured stereotypes in text-to-image outputs, but users often still prefer the stereotypical versions.

Pith tools