Fair Sampling in Diffusion Models through Switching Mechanism
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2024
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866914963036045312 |
|---|---|
| author | Choi, Yujin Park, Jinseong Kim, Hoki Lee, Jaewook Park, Saerom |
| author_facet | Choi, Yujin Park, Jinseong Kim, Hoki Lee, Jaewook Park, Saerom |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2401_03140 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Fair Sampling in Diffusion Models through Switching Mechanism Choi, Yujin Park, Jinseong Kim, Hoki Lee, Jaewook Park, Saerom Machine Learning Computers and Society 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. |
| title | Fair Sampling in Diffusion Models through Switching Mechanism |
| topic | Machine Learning Computers and Society |
| url | https://arxiv.org/abs/2401.03140 |