论文标题
对抗性潜在自动编码器
Adversarial Latent Autoencoders
论文作者
论文摘要
自动编码器网络是无监督的方法，旨在通过同时学习编码器生成器图来结合生成和代表性属性。尽管经过广泛的研究，但尚未完全解决它们是否具有相同的gan生成力量或学习分离表示的问题。我们介绍了一个共同解决这些问题的自动编码器，我们称之为对抗性潜在自动编码器（ALAE）。这是一种一般体系结构，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发电机，我们称之为styleleae。我们验证两个体系结构的分离属性。我们表明，StyLealae不仅可以生成具有可比式质量的1024x1024脸部图像，而且在相同的分辨率下还可以基于真实图像产生面部重建和操纵。这使ALAE成为第一个能够与之比较的自动编码器，并超越了仅发电机类型的体系结构的功能。
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.
