Max-Sliced Wasserstein Distance and its use for GANs
This addresses a bottleneck in GAN efficiency for high-dimensional data modeling, though it is incremental over existing sliced Wasserstein methods.
The paper tackles the high sample complexity of distance metrics in GAN training by introducing the max-sliced Wasserstein distance, which reduces projection complexity and enables training on high-dimensional images up to 256x256 resolution.
Generative adversarial nets (GANs) and variational auto-encoders have significantly improved our distribution modeling capabilities, showing promise for dataset augmentation, image-to-image translation and feature learning. However, to model high-dimensional distributions, sequential training and stacked architectures are common, increasing the number of tunable hyper-parameters as well as the training time. Nonetheless, the sample complexity of the distance metrics remains one of the factors affecting GAN training. We first show that the recently proposed sliced Wasserstein distance has compelling sample complexity properties when compared to the Wasserstein distance. To further improve the sliced Wasserstein distance we then analyze its `projection complexity' and develop the max-sliced Wasserstein distance which enjoys compelling sample complexity while reducing projection complexity, albeit necessitating a max estimation. We finally illustrate that the proposed distance trains GANs on high-dimensional images up to a resolution of 256x256 easily.