通过去除条件引导信息，让模型发现未被捕捉的潜在因素。
What We Don't C: Manifold Disentanglement for Structured Discovery
- 用隐空间流匹配显式移除条件引导信息
- 残差表示中可更易发现未被捕捉的变量
- 适合想探索模型隐藏知识的研究者
在高维数据领域，获取学习表征中的信息对标注、发现和数据筛选至关重要。我们提出 What We Don't C，一种基于隐空间流匹配的新方法，通过显式移除条件引导中包含的信息，实现隐空间的解耦，生成有意义的残差表示。这使得尚未被条件捕获的变因更加可得。我们证明了流路径中的引导会抑制引导变量的信息。结果表明，该方法是分析、控制和重用隐空间表示的简单而强大的机制，为利用生成模型探索未被捕捉、未被考虑或未被归档的内容提供了路径。
原文摘要 · Abstract (English)
Accessing information in learned representations is critical for annotation, discovery, and data filtering in disciplines where high-dimensional datasets are common. We introduce What We Don't C, a novel approach based on latent flow matching that disentangles latent subspaces by explicitly removing information included in conditional guidance, resulting in meaningful residual representations. This allows factors of variation which have not already been captured in conditioning to become more readily available. We show how guidance in the flow path necessarily represses the information from the guiding, conditioning variables. Our results highlight this approach as a simple yet powerful mechanism for analyzing, controlling, and repurposing latent representations, providing a pathway toward using generative models to explore what we don't capture, consider, or catalog.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。