# GANDiff FR: Hybrid GAN Diffusion Synthesis for Causal Bias Attribution in Face Recognition

arXiv: https://arxiv.org/abs/2508.11334 · HTML: https://mlsanity.com/paper/2508.11334/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 35 references that are in the map.
- First posted on arXiv 15 Aug 2025.

## Builds on (most cited first, 19 of 19)

- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [Alias-Free Generative Adversarial Networks](https://mlsanity.com/paper/2106.12423/) (2021)
- [CosFace: Large Margin Cosine Loss for Deep Face Recognition](https://mlsanity.com/paper/1801.09414/) (2018)
- [Nuanced Metrics for Measuring Unintended Bias with Real Data for Text Classification](https://mlsanity.com/paper/1903.04561/) (2019)
- [AdaFace: Quality Adaptive Margin for Face Recognition](https://mlsanity.com/paper/2204.00964/) (2022)
- [Towards Fairness in Visual Recognition: Effective Strategies for Bias Mitigation](https://mlsanity.com/paper/1911.11834/) (2019)
- [Compressed 3D Gaussian Splatting for Accelerated Novel View Synthesis](https://mlsanity.com/paper/2401.02436/) (2023)
- [Intrinsic Bias Metrics Do Not Correlate with Application Bias](https://mlsanity.com/paper/2012.15859/) (2020)
- [GANDiffFace: Controllable Generation of Synthetic Datasets for Face Recognition with Realistic Variations](https://mlsanity.com/paper/2305.19962/) (2023)
- [Synthetic Data for Face Recognition: Current State and Future Prospects](https://mlsanity.com/paper/2305.01021/) (2023)
- [Face Recognition Accuracy Across Demographics: Shining a Light Into the Problem](https://mlsanity.com/paper/2206.01881/) (2022)
- [Xolver: Multi-Agent Reasoning with Holistic Experience Learning Just Like an Olympiad Team](https://mlsanity.com/paper/2506.14234/) (2025)
- [Gradient-based Training of Slow Feature Analysis by Differentiable Approximate Whitening](https://mlsanity.com/paper/1808.08833/) (2018)
- [Gaussian Harmony: Attaining Fairness in Diffusion-based Face Generation Models](https://mlsanity.com/paper/2312.14976/) (2023)
- [IMVB7t: A Multi-Modal Model for Food Preferences based on Artificially Produced Traits](https://mlsanity.com/paper/2412.16807/) (2024)
- [Privacy Preserving Machine Learning Model Personalization through Federated Personalized Learning](https://mlsanity.com/paper/2505.01788/) (2025)
- [FairDiffusion: Enhancing Equity in Latent Diffusion Models via Fair Bayesian Perturbation](https://mlsanity.com/paper/2412.20374/) (2024)
- [Multilingual Question Answering in Low-Resource Settings: A Dzongkha-English Benchmark for Foundation Models](https://mlsanity.com/paper/2505.18638/) (2025)
- [Advancing Exchange Rate Forecasting: Leveraging Machine Learning and AI for Enhanced Accuracy in Global Financial Markets](https://mlsanity.com/paper/2506.09851/) (2025)

## Abstract (arXiv)

We introduce GANDiff FR, the first synthetic framework that precisely controls demographic and environmental factors to measure, explain, and reduce bias with reproducible rigor. GANDiff FR unifies StyleGAN3-based identity-preserving generation with diffusion-based attribute control, enabling fine-grained manipulation of pose around 30 degrees, illumination (four directions), and expression (five levels) under ceteris paribus conditions. We synthesize 10,000 demographically balanced faces across five cohorts validated for realism via automated detection (98.2%) and human review (89%) to isolate and quantify bias drivers. Benchmarking ArcFace, CosFace, and AdaFace under matched operating points shows AdaFace reduces inter-group TPR disparity by 60% (2.5% vs. 6.3%), with illumination accounting for 42% of residual bias. Cross-dataset evaluation on RFW, BUPT, and CASIA WebFace confirms strong synthetic-to-real transfer (r 0.85). Despite around 20% computational overhead relative to pure GANs, GANDiff FR yields three times more attribute-conditioned variants, establishing a reproducible, regulation-aligned (EU AI Act) standard for fairness auditing. Code and data are released to support transparent, scalable bias evaluation.