arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2102.06204cs.LG
收藏

非纠缠模型的非纠缠表示

Disentangled Representations from Non-Disentangled Models

Yandex(Yandex公司);Skolkovo Institute of Science and Technology (Skoltech)(斯科尔科沃科学技术研究院(Skoltech));National Research University Higher School of Economics(国立研究型大学高等经济学院)

Valentin Khrulkov, Leyla Mirvakhabova, Ivan Oseledets, Artem Babenko

摘要:

构建解纠缠表示是一项困难的任务,特别是在无监督的情况下。无监督解缠的主导范式目前是训练一个生成模型,该模型在其潜在空间中分离不同的变化因素。这种分离通常通过在模型的目标函数中使用特定的正则化项进行训练来实现。然而,这些术语引入了额外的超参数,负责解纠缠和生成质量之间的权衡。虽然调整这些超参数对于正确的解纠缠至关重要,但通常不清楚如何在没有外部监督的情况下调整它们。本文探讨了一种替代路线,以解开表示。也就是说,我们建议从最先进的生成模型中提取这样的表示,而不需要在其目标中解开术语。这种事后解纠缠的范式在学习表示时几乎不使用或不使用超参数,同时实现与现有最新技术水平相当的结果,正如通过既定解纠缠指标、公平性和抽象推理任务方面的比较所示。我们所有的代码和模型都是公开的。

英文摘要:

Constructing disentangled representations is known to be a difficult task, especially in the unsupervised scenario. The dominating paradigm of unsupervised disentanglement is currently to train a generative model that separates different factors of variation in its latent space. This separation is typically enforced by training with specific regularization terms in the model's objective function. These terms, however, introduce additional hyperparameters responsible for the trade-off between disentanglement and generation quality. While tuning these hyperparameters is crucial for proper disentanglement, it is often unclear how to tune them without external supervision. This paper investigates an alternative route to disentangled representations. Namely, we propose to extract such representations from the state-of-the-art generative models trained without disentangling terms in their objectives. This paradigm of post hoc disentanglement employs little or no hyperparameters when learning representations while achieving results on par with existing state-of-the-art, as shown by comparison in terms of established disentanglement metrics, fairness, and the abstract reasoning task. All our code and models are publicly available.

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。

↑