通过实验动作建模，实现可解释表征的严格解耦。
Disentanglement by means of action-induced representations
- 基于实验动作构建表征框架，从动作依赖性中解耦自由度。
- 提出VAIR模型，在标准VAE失效场景下实现可证明的解耦。
- 适合研究物理系统建模与因果表征的学习者。
变分自编码器（VAEs）学习可解释表征是表示学习的重要目标，核心挑战在于获得解耦表征——每个潜在维度对应一个独立生成因子。这一难题本质上源于无法进行非线性独立成分分析。本文提出动作诱导表征（AIRs）框架，通过可对物理系统执行的实验（或动作）来建模其表征。我们证明在此框架下可严格解耦与动作相关的自由度。进一步提出变分AIR架构（VAIR），能够提取AIR并实现标准VAE失败场景下的可证明解耦。除了状态表征，VAIR还捕捉生成因子的动作依赖性，直接关联实验与其影响的自由度。
原文摘要 · Abstract (English)
Learning interpretable representations with variational autoencoders (VAEs) is a major goal of representation learning. The main challenge lies in obtaining disentangled representations, where each latent dimension corresponds to a distinct generative factor. This difficulty is fundamentally tied to the inability to perform nonlinear independent component analysis. Here, we introduce the framework of action-induced representations (AIRs) which models representations of physical systems given experiments (or actions) that can be performed on them. We show that, in this framework, we can provably disentangle degrees of freedom w.r.t. their action dependence. We further introduce a variational AIR architecture (VAIR) that can extract AIRs and therefore achieve provable disentanglement where standard VAEs fail. Beyond state representation, VAIR also captures the action dependence of the underlying generative factors, directly linking experiments to the degrees of freedom they influence.
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。