What type of study is this?

This is a Quantitative Study study.

October 20, 2025Open Access

FairGen: Controlling Sensitive Attributes for Fair Generations in Diffusion Models via Adaptive Latent Guidance

View Full Paper

Ask AI

Bookmark

Authors

Proposed FairGen reduces gender bias in diffusion models, indicating enhanced ethical image generation.

FairGen achieves 68.5% reduction in gender bias during image generation, improving fairness in model outputs.
Extensive evaluations on HBE and Stable Bias datasets demonstrate superior performance over existing bias mitigation methods.
The adaptive latent guidance mechanism dynamically controls generation distribution to align with targeted attribute values.
Holistic bias evaluation benchmark HBE assesses diverse domains, ensuring comprehensive understanding of bias in diffusion models.

Kang et al. (2025) studied this question.

View Full Paper

Ask AI

Bookmark