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 2602.19461cs.CVcs.LG
收藏

拉普拉斯多尺度流匹配用于生成建模

Laplacian Multi-scale Flow Matching for Generative Modeling

Georgia Institute of Technology(佐治亚理工学院)

Zelin Zhao, Petr Molodyk, Haotian Xue, Yongxin Chen

AI总结:

本文提出拉普拉斯多尺度流匹配框架,通过并行处理多尺度表示提升图像生成质量与效率,实现高分辨率生成的同时降低计算开销。

摘要:

在本文中,我们提出了拉普拉斯多尺度流匹配(LapFlow),一种通过利用多尺度表示来增强流匹配的新型框架。我们的方法将图像分解为拉普拉斯金字塔残差,并通过具有因果注意力机制的混合转换器(MoT)架构并行处理不同尺度。与以往需要在不同尺度之间显式重新噪声的级联方法不同,我们的模型通过并行生成多尺度表示,消除了桥接过程的需要。所提出的多尺度架构不仅提高了生成质量,还加快了采样过程,并促进了扩展流匹配方法。通过在CelebA-HQ和ImageNet上的广泛实验,我们证明了我们的方法在更少的GFLOPs和更快的推理速度下,相比单尺度和多尺度流匹配基线实现了更高质量的样本生成。所提出的模型能够有效扩展到高分辨率生成(高达1024×1024)的同时保持较低的计算开销。

英文摘要:

In this paper, we present Laplacian multiscale flow matching (LapFlow), a novel framework that enhances flow matching by leveraging multi-scale representations for image generative modeling. Our approach decomposes images into Laplacian pyramid residuals and processes different scales in parallel through a mixture-of-transformers (MoT) architecture with causal attention mechanisms. Unlike previous cascaded approaches that require explicit renoising between scales, our model generates multi-scale representations in parallel, eliminating the need for bridging processes. The proposed multi-scale architecture not only improves generation quality but also accelerates the sampling process and promotes scaling flow matching methods. Through extensive experimentation on CelebA-HQ and ImageNet, we demonstrate that our method achieves superior sample quality with fewer GFLOPs and faster inference compared to single-scale and multi-scale flow matching baselines. The proposed model scales effectively to high-resolution generation (up to 1024$\times$1024) while maintaining lower computational overhead.

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。

↑