论文标题
有条件的图像生成一vs-all分类器
Conditional Image Generation with One-Vs-All Classifier
论文作者
论文摘要
本文使用基于生成对抗网络（GAN）的单VS-ALL分类器探索条件图像生成。我们建议将歧视器扩展到一个单VS-ALL分类器（GAN-OVA），而不是将每个输入数据区分为其类别标签的单VS-ALL分类器（GAN-OVA），而不是将歧视器扩展到一个单VS-ALL分类器（GAN-OVA），而不是将歧视器扩展到其类别标签的单VS-ALL分类器（GAN-OVA），而不是将歧视器扩展到其类别标签的单VS-ALL分类器。具体来说，我们将某些其他信息作为条件提供给发电机，并将鉴别器作为单VS-ALL分类器以识别每个条件类别。我们的模型可以应用于定义目标函数的不同差异或距离，例如Jensen Shannon Divergence和Earth-Mover（或称为Wasserstein-1）距离。我们在MNIST和CELEBA-HQ数据集上评估了Gan-ovas，实验结果表明，Gan-ovas在常规条件gans上取得了稳定的训练进展。此外，甘瓦斯有效地加速了不同类别的发电过程并提高了发电质量。
This paper explores conditional image generation with a One-Vs-All classifier based on the Generative Adversarial Networks (GANs). Instead of the real/fake discriminator used in vanilla GANs, we propose to extend the discriminator to a One-Vs-All classifier (GAN-OVA) that can distinguish each input data to its category label. Specifically, we feed certain additional information as conditions to the generator and take the discriminator as a One-Vs-All classifier to identify each conditional category. Our model can be applied to different divergence or distances used to define the objective function, such as Jensen-Shannon divergence and Earth-Mover (or called Wasserstein-1) distance. We evaluate GAN-OVAs on MNIST and CelebA-HQ datasets, and the experimental results show that GAN-OVAs make progress toward stable training over regular conditional GANs. Furthermore, GAN-OVAs effectively accelerate the generation process of different classes and improves generation quality.
