Sobolev Descent
This provides insight into paths to equilibrium in GANs, addressing a simplification of GAN training for researchers in generative models.
The paper tackles the problem of transporting particles from a source to a target distribution by introducing Sobolev descent, which constructs paths via gradient flows of a critic function, showing convergence to the target distribution in the MMD sense and that regularization favors smooth transitions.
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.