Saved in:
Bibliographic Details
Main Authors: Roos, Nathan, Iakovleva, Ekaterina, Gjergji, Ani, Pastore, Vito Paolo, Tartaglione, Enzo
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2512.20233
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918260938637312
author Roos, Nathan
Iakovleva, Ekaterina
Gjergji, Ani
Pastore, Vito Paolo
Tartaglione, Enzo
author_facet Roos, Nathan
Iakovleva, Ekaterina
Gjergji, Ani
Pastore, Vito Paolo
Tartaglione, Enzo
contents Diffusion-based generative models demonstrate state-of-the-art performance across various image synthesis tasks, yet their tendency to replicate and amplify dataset biases remains poorly understood. Although previous research has viewed bias amplification as an inherent characteristic of diffusion models, this work provides the first analysis of how sampling algorithms and their hyperparameters influence bias amplification. We empirically demonstrate that samplers for diffusion models -- commonly optimized for sample quality and speed -- have a significant and measurable effect on bias amplification. Through controlled studies with models trained on Biased MNIST, Multi-Color MNIST and BFFHQ, and with Stable Diffusion, we show that sampling hyperparameters can induce both bias reduction and amplification, even when the trained model is fixed. Source code is available at https://github.com/How-I-met-your-bias/how_i_met_your_bias.
format Preprint
id arxiv_https___arxiv_org_abs_2512_20233
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle How I Met Your Bias: Investigating Bias Amplification in Diffusion Models
Roos, Nathan
Iakovleva, Ekaterina
Gjergji, Ani
Pastore, Vito Paolo
Tartaglione, Enzo
Machine Learning
Computer Vision and Pattern Recognition
Diffusion-based generative models demonstrate state-of-the-art performance across various image synthesis tasks, yet their tendency to replicate and amplify dataset biases remains poorly understood. Although previous research has viewed bias amplification as an inherent characteristic of diffusion models, this work provides the first analysis of how sampling algorithms and their hyperparameters influence bias amplification. We empirically demonstrate that samplers for diffusion models -- commonly optimized for sample quality and speed -- have a significant and measurable effect on bias amplification. Through controlled studies with models trained on Biased MNIST, Multi-Color MNIST and BFFHQ, and with Stable Diffusion, we show that sampling hyperparameters can induce both bias reduction and amplification, even when the trained model is fixed. Source code is available at https://github.com/How-I-met-your-bias/how_i_met_your_bias.
title How I Met Your Bias: Investigating Bias Amplification in Diffusion Models
topic Machine Learning
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2512.20233