发表机构
University of Science and Technology of China; Hong Kong Baptist University; The Hong Kong University of Science and Technology; The University of Sydney（中国科学技术大学; 香港浸会大学; 香港科技大学; 悉尼大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究针对合成图像检测任务，发现大规模视觉模型在机器遗忘时对生成图像的特征退化更快，据此提出两种基于机器遗忘的检测方法，其性能优于传统方法，为合成图像检测建立了新范式。
AI 中文摘要
生成图像的鲁棒检测对于应对生成模型的滥用至关重要。现有方法主要依赖从人工标注的训练数据集中学习，限制了其对未见过分布的泛化能力。相比之下，在网络规模数据集上预训练的大规模视觉模型（LVMs）通过接触多样化分布展现出出色的泛化能力，为该任务提供了变革性范式。然而，我们的实验结果表明，以自然图像为主的数据预训练的LVMs能够有效捕捉自然图像和生成图像的特征，产生的损失相当低，因此在两者之间的判别能力有限。这引发了一个关键问题：LVMs在捕捉自然图像和生成图像特征时，何时以及如何表现出不同的行为？本研究揭示了一个见解：在机器遗忘过程中，LVMs表现出不同的遗忘动态，生成图像的特征退化速度比自然图像更快。受这种不同动态的启发，我们引入了两种检测方法：1）无数据检测，即修剪模型参数以在不访问数据的情况下诱导机器遗忘；2）数据驱动检测，即优化LVMs以遗忘与生成图像相关的知识。在各种基准上进行的大量实验表明，我们基于机器遗忘的方法优于传统检测方法。通过将检测任务重新定义为机器遗忘问题，本研究为生成图像检测建立了新范式。
英文摘要
Robust detection of generated images is critical to counter the misuse of generative models. Existing methods primarily depend on learning from human-annotated training datasets, limiting their generalization to unseen distributions. In contrast, large-scale vision models (LVMs) pre-trained on web-scale datasets exhibit exceptional generalization power through exposure to diverse distributions, offering a transformative paradigm for this task. However, our experimental results reveal that LVMs pre-trained on natural-image-dominated data can effectively capture the features of both natural and generated images, yielding comparably low losses and thus limited discriminative capacity between them. This prompts a key question: When and how do LVMs exhibit different behaviors when capturing features of natural and generated images? This investigation reveals an insight: during unlearning, LVMs exhibit disparate forgetting dynamics with feature degradation for generated images escalating faster than natural ones. Inspired by the disparate dynamics, we introduce two detection methods: 1) data-free detection, which prunes model parameters to induce unlearning without data access, and 2) data-driven detection, which optimizes LVMs to unlearn knowledge tied to generated images. Extensive experiments conducted on various benchmarks demonstrate that our unlearning-based approach outperforms conventional detection methods. By recasting the detection task as a problem of machine unlearning, our work establishes a new paradigm for generated image detection.
Comments19 pages, 10 figures
