Adversarial Latent Autoencoders
Adversarial Latent Autoencoders
复制标题
DOI：
10.1109/cvpr42600.2020.01411
复制
发表时间：
2020-04
期刊：
影响因子：
--
通讯作者：
Stanislav Pidhorskyi;D. Adjeroh;Gianfranco Doretto
中科院分区：
文献类型：
--
作者：
Stanislav Pidhorskyi;D. Adjeroh;Gianfranco Doretto
自动编码器网络是一种无监督的方法，旨在通过同时学习编码器-生成器映射来结合生成和表示属性。虽然被广泛研究，但它们是否具有与GAN相同的生成能力，或者学习解纠缠表示的问题尚未得到充分解决。我们引入了一个自动编码器来共同解决这些问题，我们称之为对抗性潜在自动编码器（ALAE）。它是一个通用架构，可以利用GAN培训程序的最新改进。我们设计了两个自动编码器：一个基于MLP编码器，另一个基于StyleGAN生成器，我们称之为StyleALAE。我们验证了这两种架构的解纠缠特性。我们表明，StyleALAE不仅可以生成1024 × 1024的人脸图像与StyleGAN的质量相当，但在相同的分辨率下，也可以产生人脸重建和操作的基础上真实的图像。这使得ALAE成为第一个能够与仅生成器类型的架构进行比较并超越其能力的自动编码器。
Autoencoder networks are unsupervised approaches aiming at combining generative and representational properties by learning simultaneously an encoder-generator map. Although studied extensively, the issues of whether they have the same generative power of GANs, or learn disentangled representations, have not been fully addressed. We introduce an autoencoder that tackles these issues jointly, which we call Adversarial Latent Autoencoder (ALAE). It is a general architecture that can leverage recent improvements on GAN training procedures. We designed two autoencoders: one based on a MLP encoder, and another based on a StyleGAN generator, which we call StyleALAE. We verify the disentanglement properties of both architectures. We show that StyleALAE can not only generate 1024x1024 face images with comparable quality of StyleGAN, but at the same resolution can also produce face reconstructions and manipulations based on real images. This makes ALAE the first autoencoder able to compare with, and go beyond the capabilities of a generator-only type of architecture.