几何解耦:诊断潜在空间的结构不稳定性
Geometric Decoupling: Diagnosing the Structural Instability of Latent
AI总结:
本文提出通过分析生成雅可比矩阵,将几何分解为局部缩放和局部复杂度,揭示潜在扩散模型中潜在空间不稳定性的问题,并提出几何解耦作为评估生成可靠性的新指标。
摘要:
潜在扩散模型(LDMs)能够实现高质量的合成,但其潜在空间易碎性导致编辑过程中出现不连续的语义跳跃。我们引入了一个黎曼框架,通过分析生成雅可比矩阵来诊断这种不稳定性,将几何分解为局部缩放(容量)和局部复杂度(曲率)。我们的研究表明存在一种"几何解耦":在正常生成中,曲率在功能上编码图像细节,而在分布外生成中,极端曲率被浪费在不稳定的语义边界上,而非可感知细节。这种几何误配识别出"几何热点"作为不稳定性结构的根本原因,提供了一个稳健的内在度量标准来诊断生成可靠性。
英文摘要:
Latent Diffusion Models (LDMs) achieve high-fidelity synthesis but suffer from latent space brittleness, causing discontinuous semantic jumps during editing. We introduce a Riemannian framework to diagnose this instability by analyzing the generative Jacobian, decomposing geometry into \textit{Local Scaling} (capacity) and \textit{Local Complexity} (curvature). Our study uncovers a \textbf{``Geometric Decoupling"}: while curvature in normal generation functionally encodes image detail, OOD generation exhibits a functional decoupling where extreme curvature is wasted on unstable semantic boundaries rather than perceptible details. This geometric misallocation identifies ``Geometric Hotspots" as the structural root of instability, providing a robust intrinsic metric for diagnosing generative reliability.
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。