Generalized Loss-Sensitive Adversarial Learning with Manifold Margins
Generalized Loss-Sensitive Adversarial Learning with Manifold Margins
复制标题
DOI：
10.1007/978-3-030-01228-1_6
复制
发表时间：
2018-09
期刊：
影响因子：
5.2
通讯作者：
Marzieh Edraki;Guo-Jun Qi
中科院分区：
文献类型：
--
作者：
Marzieh Edraki;Guo-Jun Qi
经典的生成对抗网络及其变体可以大致分为两大类：非正则化与正则化GAN。通过放松经典GAN中的非参数假设，正则化GAN具有更好的泛化能力，可以从真实的分布中产生新的样本。众所周知，像自然图像这样的真实的数据在整个数据空间上不是均匀分布的。相反，它们通常被限制在周围空间的低维流形上。这种流形假设表明，流形上的距离应该是表征真实的和伪样本之间差异的更好度量。因此，我们定义了一个回调算子将样本映射回它们的数据流形，并将流形边缘定义为回调表示之间的距离，以区分真实的和假样本，并学习最佳生成器。我们证明了所提出的模型的有效性，从理论和经验。
The classic Generative Adversarial Net and its variants can be roughly categorized into two large families: the unregularized ver-sus regularized GANs. By relaxing the non-parametric assumption on the discriminator in the classic GAN, the regularized GANs have better generalization ability to produce new samples drawn from the real dis-tribution. It is well known that the real data like natural images are not uniformly distributed over the whole data space. Instead, they are often restricted to a low-dimensional manifold of the ambient space. Such a manifold assumption suggests the distance over the manifold should be a better measure to characterize the distinct between real and fake sam-ples. Thus, we define a pullback operator to map samples back to their data manifold, and a manifold margin is defined as the distance between the pullback representations to distinguish between real and fake sam-ples and learn the optimal generators. We justify the effectiveness of the proposed model both theoretically and empirically.