我们未捕捉到的:结构发现的流解耦
What We Don't C: Manifold Disentanglement for Structured Discovery
AI总结:
What We Don't C通过移除条件指导中的信息来解耦潜在子空间,从而提升对潜在表示的分析与再利用能力。
摘要:
在高维数据集常见的学科中,访问学习到的表示中的信息对于注释、发现和数据过滤至关重要。我们引入了What We Don't C,一种基于潜在流匹配的新型方法,通过显式地移除条件指导中包含的信息来解耦潜在子空间,从而得到有意义的残差表示。这使得那些尚未在条件中被捕捉到的变化因素变得更加容易获得。我们展示了流路径中的指导必然会压制来自指导和条件变量的信息。我们的结果突显了这种方法作为分析、控制和重新利用潜在表示的简单而有力的机制,为使用生成模型探索我们未捕捉、考虑或记录的内容提供了途径。
英文摘要:
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.
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。