用马尔可夫毯掩盖CelebA-like数据集中的偏差:属性先验避免的事后治愈
Covering up bias in CelebA-like datasets with Markov blankets: A post-hoc cure for attribute prior avoidance
摘要:
属性先验避免需要对数据集通常与生俱来的(Meta)属性进行潜意识或故意的非建模,例如与CelebA和CelebA-HQ数据集相关联的40个语义面部属性。我们发现,这种弱点的后果在这些数据集上学习的最先进的深度生成模型中尤为明显,这些模型只是对像素空间测量进行建模,从而导致了一个充满属性间偏见的潜在空间。当我们进行基于潜在向量插值的人脸操作实验时,这一点在内心表现出来。在本文中,我们解决了这个问题,并提出了一个事后解决方案,利用伊辛属性事先学习的属性空间,并通过定性实验展示其功效。
英文摘要:
Attribute prior avoidance entails subconscious or willful non-modeling of (meta)attributes that datasets are oft born with, such as the 40 semantic facial attributes associated with the CelebA and CelebA-HQ datasets. The consequences of this infirmity, we discover, are especially stark in state-of-the-art deep generative models learned on these datasets that just model the pixel-space measurements, resulting in an inter-attribute bias-laden latent space. This viscerally manifests itself when we perform face manipulation experiments based on latent vector interpolations. In this paper, we address this and propose a post-hoc solution that utilizes an Ising attribute prior learned in the attribute space and showcase its efficacy via qualitative experiments.
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。