

Preprint 13/2019
Wasserstein of Wasserstein Loss for Learning Generative Models
Yonatan Dukler, Wuchen Li, Alex Tong Lin, and Guido Montúfar
Contact the author: Please use for correspondence this email.
Submission date: 29. Jan. 2019
Pages: 24
published in: Proceedings of the 36th international conference on machine learning, 9-15 June 2019, Long Beach, California, USA / K. Chaudhuri (ed.)
Long Beach, California : PMLR, 2019. - P. 1716 - 1725
(Proceedings of machine learning research ; 97)
Bibtex
Keywords and phrases: Wasserstein metric, GAN, statistical manifold
Download full preprint: PDF (1145 kB)
Abstract:
The Wasserstein distance serves as a loss function for unsupervised learning which depends on the choice of a ground metric on sample space. We propose to use a Wasserstein distance as the ground metric on the sample space of images. This ground metric is known as an effective distance for image retrieval, since it correlates with human perception. We derive the Wasserstein ground metric on image space and define a Riemannian Wasserstein gradient penalty to be used in the Wasserstein Generative Adversarial Network (WGAN) framework. The new gradient penalty is computed efficiently via convolutions on the L2 (Euclidean) gradients with negligible additional computational cost. The new formulation is more robust to the natural variability of images and provides for a more continuous discriminator in sample space.