Title: Improved BiGAN training with marginal likelihood equalization Show Chinese title

Title: 通过边缘似然均衡改进BiGAN训练

Abstract: We propose a novel training procedure for improving the performance of generative adversarial networks (GANs), especially to bidirectional GANs. First, we enforce that the empirical distribution of the inverse inference network matches the prior distribution, which favors the generator network reproducibility on the seen samples. Second, we have found that the marginal log-likelihood of the samples shows a severe overrepresentation of a certain type of samples. To address this issue, we propose to train the bidirectional GAN using a non-uniform sampling for the mini-batch selection, resulting in improved quality and variety in generated samples measured quantitatively and by visual inspection. We illustrate our new procedure with the well-known CIFAR10, Fashion MNIST and CelebA datasets. Show Chinese abstract

Abstract: 我们提出了一种新颖的训练过程，以提高生成对抗网络（GANs）的性能，特别是针对双向GANs。 首先，我们强制逆向推理网络的经验分布与先验分布匹配，这有利于生成器网络在已见样本上的可重复性。 其次，我们发现样本的边缘对数似然严重地高估了某种类型的样本。 为了解决这个问题，我们提出使用非均匀采样进行小批量选择来训练双向GAN，从而在定量测量和视觉检查中提高了生成样本的质量和多样性。 我们使用著名的CIFAR10、Fashion MNIST和CelebA数据集来说明我们的新过程。