为扩散模型生成样本的质量不确定性提供可落地的贝叶斯评估方法
Generative Uncertainty in Diffusion Models
- 基于贝叶斯框架，利用潜在空间语义似然估计生成样本的不确定性
- 能有效识别低质量样本，性能显著优于现有不确定性方法
- 可后处理任意预训练扩散或流匹配模型，计算开销极低
扩散模型在生成建模中取得了显著突破。尽管顶尖模型平均生成质量高，但个别样本仍可能质量低下。如何在无需人工标注的情况下检测此类样本仍是难题。为此，我们提出一种贝叶斯框架，用于估计合成样本的生成不确定性。我们阐述了如何将贝叶斯推断应用于大型现代生成模型，并引入一种新的语义似然（在特征提取器的潜在空间中评估），以应对高维样本空间带来的挑战。实验表明，所提出的生成不确定性能够有效识别低质量样本，且显著优于现有基于不确定性的方法。值得注意的是，我们的贝叶斯框架可后处理任意预训练的扩散模型或流匹配模型（通过拉普拉斯近似），并提出了简单而有效的技术以最小化采样时的计算开销。
原文摘要 · Abstract (English)
Diffusion models have recently driven significant breakthroughs in generative modeling. While state-of-the-art models produce high-quality samples on average, individual samples can still be low quality. Detecting such samples without human inspection remains a challenging task. To address this, we propose a Bayesian framework for estimating generative uncertainty of synthetic samples. We outline how to make Bayesian inference practical for large, modern generative models and introduce a new semantic likelihood (evaluated in the latent space of a feature extractor) to address the challenges posed by high-dimensional sample spaces. Through our experiments, we demonstrate that the proposed generative uncertainty effectively identifies poor-quality samples and significantly outperforms existing uncertainty-based methods. Notably, our Bayesian framework can be applied post-hoc to any pretrained diffusion or flow matching model (via the Laplace approximation), and we propose simple yet effective techniques to minimize its computational overhead during sampling.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。