Sobolev Descent

TL;DR

This paper introduces Sobolev descent, a method for transporting particles from a source to a target distribution using gradient flows, linking it to optimal transport and improving distribution transition smoothness.

Contribution

It presents Sobolev descent as a novel particle transport method related to optimal transport, with theoretical convergence guarantees and practical benefits in smooth distribution transitions.

Findings

Kernel version converges to target in MMD sense

Regularization promotes smooth distribution transitions

Provides insight into GAN equilibrium paths

Abstract

We study a simplification of GAN training: the problem of transporting particles from a source to a target distribution. Starting from the Sobolev GAN critic, part of the gradient regularized GAN family, we show a strong relation with Optimal Transport (OT). Specifically with the less popular dynamic formulation of OT that finds a path of distributions from source to target minimizing a ``kinetic energy''. We introduce Sobolev descent that constructs similar paths by following gradient flows of a critic function in a kernel space or parametrized by a neural network. In the kernel version, we show convergence to the target distribution in the MMD sense. We show in theory and experiments that regularization has an important role in favoring smooth transitions between distributions, avoiding large gradients from the critic. This analysis in a simplified particle setting provides insight in paths to equilibrium in GANs.