Scholarlink

<p>Diffusion models have demonstrated remarkable success in generating diverse and high-fidelity images. However, their performance often degrades when applied to long-tailed datasets, where head classes significantly outnumber tail classes. This imbalance leads to biased model training, favoring head classes and neglecting tail classes. To address this challenge, we conduct an in-depth analysis of the Lipschitz singularity problem that occurs when the time-steps are near zero in long-tailed diffusion models, which leads to numerical instability and degraded image quality. We propose a time-step sharing-based class-balancing diffusion model (TCDM) that effectively mitigates this issue by combining a shared time-step strategy with a conditional probability transfer mechanism. TCDM improves noise prediction accuracy and information transfer stability, leading to enhanced image generation quality for tail classes. Experimental results on four long-tailed datasets, including CIFAR-100LT, CIFAR-10LT, TinyImageNet-200LT, and CelebA-5, demonstrate TCDM’s superior performance, achieving leading FID scores and higher recall rates compared to existing methods. Here, we show that TCDM significantly reduces the Lipschitz constant when the time-steps are near zero, thereby ensuring more stable and accurate noise prediction and feature transfer. This research contributes to the broader field of generative models by providing a robust solution for handling long-tailed distributions in image generation tasks. Our code is available at <a href="https://github.com/shiyanbei306/TCDM">https://github.com/shiyanbei306/TCDM</a>.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Mitigating Lipschitz singularities in long-tailed diffusion models via time-step sharing strategy

  • 期刊论文

  • 发表日期：

    2025年12月17日

查看全文
  • Qiangkui Leng,
  • Zhuoyu Zhou,
  • Keyi Song,
  • Guansheng Yuan

摘要

Diffusion models have demonstrated remarkable success in generating diverse and high-fidelity images. However, their performance often degrades when applied to long-tailed datasets, where head classes significantly outnumber tail classes. This imbalance leads to biased model training, favoring head classes and neglecting tail classes. To address this challenge, we conduct an in-depth analysis of the Lipschitz singularity problem that occurs when the time-steps are near zero in long-tailed diffusion models, which leads to numerical instability and degraded image quality. We propose a time-step sharing-based class-balancing diffusion model (TCDM) that effectively mitigates this issue by combining a shared time-step strategy with a conditional probability transfer mechanism. TCDM improves noise prediction accuracy and information transfer stability, leading to enhanced image generation quality for tail classes. Experimental results on four long-tailed datasets, including CIFAR-100LT, CIFAR-10LT, TinyImageNet-200LT, and CelebA-5, demonstrate TCDM’s superior performance, achieving leading FID scores and higher recall rates compared to existing methods. Here, we show that TCDM significantly reduces the Lipschitz constant when the time-steps are near zero, thereby ensuring more stable and accurate noise prediction and feature transfer. This research contributes to the broader field of generative models by providing a robust solution for handling long-tailed distributions in image generation tasks. Our code is available at https://github.com/shiyanbei306/TCDM.