HoloFair: Unified T2I Fairness Evaluation and Fair-GRPO Debiasing

• URL: http://arxiv.org/abs/2605.24687v1
• Date: Sat, 23 May 2026 17:52:07 GMT
• Title: HoloFair: Unified T2I Fairness Evaluation and Fair-GRPO Debiasing
• Abstract summary: Existing evaluation methods address only single-dimensional biases.<n>We introduce HoloFair, a comprehensive benchmark framework for multidimensional demographic bias analysis.<n>We also introduce Fair-GRPO, a reinforcement-learning-based debiasing method.
• Score: 20.795940054868854
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Text-to-Image (T2I) models have made significant strides in visual realism and semantic consistency, yet they often perpetuate and amplify societal biases. Existing evaluation methods typically address only single-dimensional biases, lacking perspectives to uncover model biases at social-related deeper semantic levels. We introduce HoloFair, a comprehensive benchmark framework for multidimensional demographic bias analysis. Built upon our large-scale fairness-oriented dataset and the SpaFreq (Spatial-Frequency) attribute classifier, this framework proposes the Multi-attribute, Group-wise Bias Index (MGBI) metric, designed to assess both intrinsic diversity and conditional biases. Beyond evaluation, we further introduce Fair-GRPO, a reinforcement-learning-based debiasing method that alters the distribution of generative models through a designed multi-objective reward function. E.g., experiments on the SD3.5-Medium model demonstrate that Fair-GRPO significantly improves multidimensional fairness while maintaining high image quality. We also analyze potential reward hacking phenomena and provide corresponding mitigation strategies. Code and dataset are available at https://github.com/1059684669/HoloFair

Related papers

• BiasIG: Benchmarking Multi-dimensional Social Biases in Text-to-Image Models [30.463069540851688]
  We introduce BiasIG, a unified benchmark that quantifies social biases across a curated dataset of 47,040 prompts.<n>Extensive experiments on 8 T2I models and 3 debiasing methods validate BiasIG as a robust diagnostic tool.<n>Our work advocates for a precise, taxonomy-driven approach to fairness in AIGC, providing a theoretical framework for using BiasIG's metrics as feedback signals in future closed-loop mitigation.
  arXiv  Detail & Related papers  (2026-04-13T18:22:50Z)
• Understanding trade-offs in classifier bias with quality-diversity optimization: an application to talent management [2.334978724544296]
  A major struggle for the development of fair AI models lies in the bias implicit in the data available to train such models.<n>We propose a method for visualizing the biases inherent in a dataset and understanding the potential trade-offs between fairness and accuracy.
  arXiv  Detail & Related papers  (2024-11-25T22:14:02Z)
• On the Fairness, Diversity and Reliability of Text-to-Image Generative Models [68.62012304574012]
  multimodal generative models have sparked critical discussions on their reliability, fairness and potential for misuse.<n>We propose an evaluation framework to assess model reliability by analyzing responses to global and local perturbations in the embedding space.<n>Our method lays the groundwork for detecting unreliable, bias-injected models and tracing the provenance of embedded biases.
  arXiv  Detail & Related papers  (2024-11-21T09:46:55Z)
• Bias Begets Bias: The Impact of Biased Embeddings on Diffusion Models [0.0]
  Text-to-Image (TTI) systems have come under increased scrutiny for social biases. We investigate embedding spaces as a source of bias for TTI models. We find that biased multimodal embeddings like CLIP can result in lower alignment scores for representationally balanced TTI models.
  arXiv  Detail & Related papers  (2024-09-15T01:09:55Z)
• Identifying and Mitigating Social Bias Knowledge in Language Models [52.52955281662332]
  We propose a novel debiasing approach, Fairness Stamp (FAST), which enables fine-grained calibration of individual social biases.<n>FAST surpasses state-of-the-art baselines with superior debiasing performance.<n>This highlights the potential of fine-grained debiasing strategies to achieve fairness in large language models.
  arXiv  Detail & Related papers  (2024-08-07T17:14:58Z)
• Thinking Racial Bias in Fair Forgery Detection: Models, Datasets and Evaluations [63.52709761339949]
  We first contribute a dedicated dataset called the Fair Forgery Detection (FairFD) dataset, where we prove the racial bias of public state-of-the-art (SOTA) methods.<n>We design novel metrics including Approach Averaged Metric and Utility Regularized Metric, which can avoid deceptive results.<n>We also present an effective and robust post-processing technique, Bias Pruning with Fair Activations (BPFA), which improves fairness without requiring retraining or weight updates.
  arXiv  Detail & Related papers  (2024-07-19T14:53:18Z)
• Evaluating the Fairness of Discriminative Foundation Models in Computer Vision [51.176061115977774]
  We propose a novel taxonomy for bias evaluation of discriminative foundation models, such as Contrastive Language-Pretraining (CLIP) We then systematically evaluate existing methods for mitigating bias in these models with respect to our taxonomy. Specifically, we evaluate OpenAI's CLIP and OpenCLIP models for key applications, such as zero-shot classification, image retrieval and image captioning.
  arXiv  Detail & Related papers  (2023-10-18T10:32:39Z)
• Fairness meets Cross-Domain Learning: a new perspective on Models and Metrics [80.07271410743806]
  We study the relationship between cross-domain learning (CD) and model fairness. We introduce a benchmark on face and medical images spanning several demographic groups as well as classification and localization tasks. Our study covers 14 CD approaches alongside three state-of-the-art fairness algorithms and shows how the former can outperform the latter.
  arXiv  Detail & Related papers  (2023-03-25T09:34:05Z)
• DualFair: Fair Representation Learning at Both Group and Individual Levels via Contrastive Self-supervision [73.80009454050858]
  This work presents a self-supervised model, called DualFair, that can debias sensitive attributes like gender and race from learned representations. Our model jointly optimize for two fairness criteria - group fairness and counterfactual fairness.
  arXiv  Detail & Related papers  (2023-03-15T07:13:54Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.