揭示扩散模型泛化机制：局部去噪是关键。
Towards a Mechanistic Explanation of Diffusion Model Generalization
- 通过对比神经网络与理论最优去噪器，发现共享的局部归纳偏置。
- 提出的新算法在去噪过程中的视觉相似度更高，均方误差更低。
- 适合对扩散模型原理感兴趣的研究人员和深度学习从业者。
我们提出一种简单且无需训练的机制，解释扩散模型的泛化行为。通过比较预训练扩散模型与其理论上最优的经验对应物，我们发现多种网络架构间存在共享的局部归纳偏置。基于此，我们假设网络去噪器通过局部去噪操作实现泛化，因为这些操作在训练分布的大部分区域能很好地逼近训练目标。为验证该假设，我们引入新的去噪算法，聚合局部经验去噪器以复现网络行为。在正向与逆向扩散过程中，将该方法与网络去噪器进行比较，结果表明其输出在视觉上与神经网络输出具有一致性，且均方误差低于以往方法。
原文摘要 · Abstract (English)
We propose a simple, training-free mechanism which explains the generalization behaviour of diffusion models. By comparing pre-trained diffusion models to their theoretically optimal empirical counterparts, we identify a shared local inductive bias across a variety of network architectures. From this observation, we hypothesize that network denoisers generalize through localized denoising operations, as these operations approximate the training objective well over much of the training distribution. To validate our hypothesis, we introduce novel denoising algorithms which aggregate local empirical denoisers to replicate network behaviour. Comparing these algorithms to network denoisers across forward and reverse diffusion processes, our approach exhibits consistent visual similarity to neural network outputs, with lower mean squared error than previously proposed methods.
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。