AI 中文总结
本研究比较扩散采样中早期扰动与新鲜噪声对晚期状态的影响，发现早期方向在角度和特征层面增益更大，且依赖轨迹对齐。核心方法是跨时间传输扰动并重放方向，主要贡献是揭示方向选择效应。
AI 中文摘要
剩余的扩散采样步骤会在多大程度上放大晚期潜在状态下的扰动？标准测量方法使用新采样的各向同性噪声来回答这个问题，尽管采样过程中遇到的扰动已被早期步骤变换。我们直接比较这两种情况。对于每条轨迹，我们将一个居中的扰动从早期步骤传输到晚期状态，然后以与新采样的各向同性扰动相同的幅度重放其方向；两者随后经历相同的剩余更新。在我们研究的采样器中，成对的有形状与新鲜的角度增益比的中位数范围从1.14到2.35。在原始主要队列中，每条轨迹的有形状角度增益都超过其匹配的新鲜对应物。当端点变化通过潜在RMS测量时，也出现同样的模式。该效应也出现在感知特征表示中：早期采样方向导致端点处的特征变化更大，即使对应的像素空间变化相当。跨轨迹置换有形状方向会削弱该效应，包括在类别内，表明该优势依赖于与接收轨迹的对齐以及共享的方向结构。
英文摘要
How strongly do the remaining diffusion-sampling steps amplify a perturbation at a late latent state? Standard measurements answer this question with newly sampled isotropic noise, even though perturbations encountered during sampling have already been transformed by earlier steps. We compare these two cases directly. For each trajectory, we transport a centered perturbation from an earlier step to a late state, then replay its direction at the same magnitude as a newly sampled isotropic perturbation; both then undergo the same remaining updates. Across the samplers we study, median paired shaped-to-fresh angular-gain ratios range from $1.14$ to $2.35$. Shaped angular gain exceeds its matched fresh counterpart in every trajectory in the original main cohorts. The same pattern appears when endpoint change is measured by latent RMS. The effect also appears in perceptual feature representations: earlier-sampling directions cause larger feature changes at the endpoint, even when the corresponding pixel-space change is comparable. Permuting shaped directions across trajectories weakens the effect, including within class, indicating that the advantage depends on alignment with the receiving trajectory as well as on shared directional structure.
Comments12 pages, 6 figures. NeurIPS 2026 AI for Stochastic Dynamics Workshop (poster)
