IVE-GAN：不变编码生成对抗网络

Nov, 2017

IVE-GAN: Invariant Encoding Generative Adversarial Networks

Robin Winter, Djork-Arné Clevert

TL;DR提出了一种新的“不变编码生成式对抗网络”（IVE-GANs）框架，利用特定数据的不变特征进行逆向映射，使生成器可以涵盖更多数据模式并提升生成效果和表示能力，并在多个数据集上进行了验证。

Abstract

generative adversarial networks (GANs) are a powerful framework for generative tasks. However, they are difficult to train and tend to miss modes of the true data generation process. Although GANs can learn a ric