Fair Sampling in Diffusion Models through Switching Mechanism
DOI:
https://doi.org/10.1609/aaai.v38i20.30202Keywords:
GeneralAbstract
Diffusion models have shown their effectiveness in generation tasks by well-approximating the underlying probability distribution. However, diffusion models are known to suffer from an amplified inherent bias from the training data in terms of fairness. While the sampling process of diffusion models can be controlled by conditional guidance, previous works have attempted to find empirical guidance to achieve quantitative fairness. To address this limitation, we propose a fairness-aware sampling method called \textit{attribute switching} mechanism for diffusion models. Without additional training, the proposed sampling can obfuscate sensitive attributes in generated data without relying on classifiers. We mathematically prove and experimentally demonstrate the effectiveness of the proposed method on two key aspects: (i) the generation of fair data and (ii) the preservation of the utility of the generated data.Downloads
Published
2024-03-24
How to Cite
Choi, Y., Park, J., Kim, H., Lee, J., & Park, S. (2024). Fair Sampling in Diffusion Models through Switching Mechanism. Proceedings of the AAAI Conference on Artificial Intelligence, 38(20), 21995-22003. https://doi.org/10.1609/aaai.v38i20.30202
Issue
Section
AAAI Technical Track on AI for Social Impact Track