论文标题
多层潜在空间结构用于生成控制
Multi-level Latent Space Structuring for Generative Control
论文作者
论文摘要
截断是在生成模型中广泛用于提高生成样品质量的，以降低其多样性为代价。我们建议利用Stylegan生成架构来设计一种新的截断技术，该技术基于将潜在空间的分解为群集，从而可以在多个语义级别执行自定义的截断。我们通过学习重新生成W-Space（使用高斯人的可学习混合物）重新生成W-Space，同时训练分类器，以识别每个潜在向量的群集。由此产生的截断方案更忠实于原始的未截断样品，并可以在质量和多样性之间进行更好的权衡。我们将我们的方法与其他截断方法进行比较，无论是定性和定量的。
Truncation is widely used in generative models for improving the quality of the generated samples, at the expense of reducing their diversity. We propose to leverage the StyleGAN generative architecture to devise a new truncation technique, based on a decomposition of the latent space into clusters, enabling customized truncation to be performed at multiple semantic levels. We do so by learning to re-generate W-space, the extended intermediate latent space of StyleGAN, using a learnable mixture of Gaussians, while simultaneously training a classifier to identify, for each latent vector, the cluster that it belongs to. The resulting truncation scheme is more faithful to the original untruncated samples and allows a better trade-off between quality and diversity. We compare our method to other truncation approaches for StyleGAN, both qualitatively and quantitatively.
