扩散心智平均
Diffusion Mental Averages
把「求平均」从像素空间搬到扩散模型的语义空间，用轨迹对齐优化多个噪声潜变量使其去噪轨迹收敛到同一原型——这提供了一个无需解码器即可在扩散潜空间做语义聚合/编辑的可复用手段，也可用于探测模型偏见与概念表征。
原文依据 we cast averaging as trajectory alignment: optimize multiple noise latents so their denoising trajectories progressively converge
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论扩散模型能否为一个概念生成它自己的「心智平均」——一个和典型样本一样清晰、真实的平均？我们提出扩散心智平均（DMA），从模型中心的角度回答这一问题。此前的方法旨在对图像集合求平均，但当它们被用于同一提示词产生的扩散样本时，结果会变得模糊。这些数据中心的技术在模型之外运作，忽略了生成过程。相比之下，DMA 在扩散模型的语义空间内求平均，该空间已被近期研究揭示。由于这一空间随时间步演化且缺少直接解码器，我们将求平均表述为轨迹对齐：优化多个噪声潜变量，使其去噪轨迹逐步收敛到共享的由粗到细的语义，从而得到单个清晰的原型。我们通过在以 CLIP 等语义丰富的空间中聚类样本，并应用文本反转或低秩适配把 CLIP 簇桥接到扩散空间，将该方法扩展到多模态概念（例如包含众多品种的狗）。据我们所知，这是首个能给出连贯、真实平均结果（即使对抽象概念也如此）的方法，它既可作为具体的视觉摘要，也可作为观察模型偏见与概念表征的透镜。
英文原文摘要 1256 字符
Can a diffusion model produce its own "mental average" of a concept--one that is as sharp and realistic as a typical sample? We introduce Diffusion Mental Averages (DMA), a model-centric answer to this question. While prior methods aim to average image collections, they produce blurry results when applied to diffusion samples from the same prompt. These data-centric techniques operate outside the model, ignoring the generative process. In contrast, DMA averages within the diffusion model's semantic space, as discovered by recent studies. Since this space evolves across timesteps and lacks a direct decoder, we cast averaging as trajectory alignment: optimize multiple noise latents so their denoising trajectories progressively converge toward shared coarse-to-fine semantics, yielding a single sharp prototype. We extend our approach to multimodal concepts (e.g., dogs with many breeds) by clustering samples in semantically-rich spaces such as CLIP and applying Textual Inversion or LoRA to bridge CLIP clusters into diffusion space. This is, to our knowledge, the first approach that delivers consistent, realistic averages, even for abstract concepts, serving as a concrete visual summary and a lens into model biases and concept representation.