ADT: Tuning Diffusion Models with Adversarial Supervision

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Shen, Dazhong, Song, Guanglu, Zhang, Yi, Ma, Bingqi, Li, Lujundong, Jiang, Dongzhi, Zong, Zhuofan, Liu, Yu
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916691818053632
author Shen, Dazhong
Song, Guanglu
Zhang, Yi
Ma, Bingqi
Li, Lujundong
Jiang, Dongzhi
Zong, Zhuofan
Liu, Yu
author_facet Shen, Dazhong
Song, Guanglu
Zhang, Yi
Ma, Bingqi
Li, Lujundong
Jiang, Dongzhi
Zong, Zhuofan
Liu, Yu
contents Diffusion models have achieved outstanding image generation by reversing a forward noising process to approximate true data distributions. During training, these models predict diffusion scores from noised versions of true samples in a single forward pass, while inference requires iterative denoising starting from white noise. This training-inference divergences hinder the alignment between inference and training data distributions, due to potential prediction biases and cumulative error accumulation. To address this problem, we propose an intuitive but effective fine-tuning framework, called Adversarial Diffusion Tuning (ADT), by stimulating the inference process during optimization and aligning the final outputs with training data by adversarial supervision. Specifically, to achieve robust adversarial training, ADT features a siamese-network discriminator with a fixed pre-trained backbone and lightweight trainable parameters, incorporates an image-to-image sampling strategy to smooth discriminative difficulties, and preserves the original diffusion loss to prevent discriminator hacking. In addition, we carefully constrain the backward-flowing path for back-propagating gradients along the inference path without incurring memory overload or gradient explosion. Finally, extensive experiments on Stable Diffusion models (v1.5, XL, and v3), demonstrate that ADT significantly improves both distribution alignment and image quality.
format Preprint
id arxiv_https___arxiv_org_abs_2504_11423
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ADT: Tuning Diffusion Models with Adversarial Supervision
Shen, Dazhong
Song, Guanglu
Zhang, Yi
Ma, Bingqi
Li, Lujundong
Jiang, Dongzhi
Zong, Zhuofan
Liu, Yu
Computer Vision and Pattern Recognition
Artificial Intelligence
Diffusion models have achieved outstanding image generation by reversing a forward noising process to approximate true data distributions. During training, these models predict diffusion scores from noised versions of true samples in a single forward pass, while inference requires iterative denoising starting from white noise. This training-inference divergences hinder the alignment between inference and training data distributions, due to potential prediction biases and cumulative error accumulation. To address this problem, we propose an intuitive but effective fine-tuning framework, called Adversarial Diffusion Tuning (ADT), by stimulating the inference process during optimization and aligning the final outputs with training data by adversarial supervision. Specifically, to achieve robust adversarial training, ADT features a siamese-network discriminator with a fixed pre-trained backbone and lightweight trainable parameters, incorporates an image-to-image sampling strategy to smooth discriminative difficulties, and preserves the original diffusion loss to prevent discriminator hacking. In addition, we carefully constrain the backward-flowing path for back-propagating gradients along the inference path without incurring memory overload or gradient explosion. Finally, extensive experiments on Stable Diffusion models (v1.5, XL, and v3), demonstrate that ADT significantly improves both distribution alignment and image quality.
title ADT: Tuning Diffusion Models with Adversarial Supervision
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2504.11423