arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2510.22196cs.CVcs.AI
收藏

通过表示实现非参数采样的扩展

Scaling Non-Parametric Sampling with Representation

University of Wisconsin–Madison(威斯康星大学麦迪逊分校);University of Illinois Urbana–Champaign(伊利诺伊大学厄巴纳-香槟分校);UC Berkeley(伯克利大学);UC Davis(戴维斯大学)

Vincent Lu, Aaron Truong, Zeyu Yun, Yubei Chen

AI总结:

本文提出一个无需训练的非参数生成模型,基于自然图像三原则定义像素分布,在MNIST和CIFAR-10上表现优异,并通过白盒分析揭示部分-整体泛化的组合机制。

摘要:

扩展和架构上的进步已经产生了令人惊叹的照片级逼真图像生成模型,但其机制仍然不透明。我们的目标不是推进扩展,而是剥离复杂的工程技巧,提出一个简单的非参数生成模型。我们的设计基于自然图像的三个原则——(i)空间非平稳性,(ii)低级规律性,以及(iii)高级语义——并定义了每个像素在其局部上下文窗口中的分布。尽管其架构极简且无需训练,该模型在MNIST上生成了高保真样本,并在CIFAR-10上生成了视觉上引人注目的图像。这种简单性与强大实证表现的结合指向了一个关于自然图像结构的最小理论。该模型的白盒特性还使我们能够从机制上理解模型如何泛化并生成多样化的图像。我们通过将每个生成的像素追溯到其源图像来研究这一点。这些分析揭示了一个简单的“部分-整体泛化”组合过程,为大型神经网络生成模型如何学习泛化提出了一个假设。

英文摘要:

Scaling and architectural advances have produced strikingly photorealistic image generative models, yet their mechanisms still remain opaque. Rather than advancing scaling, our goal is to strip away complicated engineering tricks and propose a simple, non-parametric generative model. Our design is grounded in three principles of natural images-(i) spatial non-stationarity, (ii) low-level regularities, and (iii) high-level semantics-and defines each pixel's distribution from its local context window. Despite its minimal architecture and no training, the model produces high-fidelity samples on MNIST and visually compelling CIFAR-10 images. This combination of simplicity and strong empirical performance points toward a minimal theory of natural-image structure. The model's white-box nature also allows us to have a mechanistic understanding of how the model generalizes and generates diverse images. We study it by tracing each generated pixel back to its source images. These analyses reveal a simple, compositional procedure for "part-whole generalization", suggesting a hypothesis for how large neural network generative models learn to generalize.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑