arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2103.01678stat.MLcs.LG
收藏

Wasserstein GANs工作是因为它们失败了(近似瓦瑟斯坦距离)

Wasserstein GANs Work Because They Fail (to Approximate the Wasserstein Distance)

Cambridge Image Analysis Group, Department of Applied Mathematics and Theoretical Physics, University of Cambridge(剑桥大学应用数学与理论物理系剑桥图像分析组);Department of Mathematical Sciences, University of Bath(巴斯大学数学科学系);University of Cambridge(剑桥大学);University of Bath(巴斯大学)

Jan Stanczuk, Christian Etmann, Lisa Maria Kreusser, Carola-Bibiane Schönlieb

摘要:

Wasserstein GANs基于最小化真实分布和生成分布之间的瓦瑟斯坦距离的想法。我们对训练Wasserstein GANs的理论设置和现实之间的差异进行了深入的数学分析。在这项工作中,我们收集的理论和经验证据表明,WGAN的损失是没有意义的近似的瓦瑟斯坦距离。此外,我们认为Wasserstein距离甚至不是深度生成模型的理想损失函数,并得出结论,Wasserstein GANs的成功实际上可以归因于未能近似瓦瑟斯坦距离。

英文摘要:

Wasserstein GANs are based on the idea of minimising the Wasserstein distance between a real and a generated distribution. We provide an in-depth mathematical analysis of differences between the theoretical setup and the reality of training Wasserstein GANs. In this work, we gather both theoretical and empirical evidence that the WGAN loss is not a meaningful approximation of the Wasserstein distance. Moreover, we argue that the Wasserstein distance is not even a desirable loss function for deep generative models, and conclude that the success of Wasserstein GANs can in truth be attributed to a failure to approximate the Wasserstein distance.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑