Skip to yearly menu bar Skip to main content


Learning disconnected manifolds: a no GAN's land

Ugo Tanielian · Thibaut Issenhuth · Elvis Dohmatob · Jeremie Mary

Keywords: [ Generative Adversarial Networks ] [ Deep Learning - General ]


Typical architectures of Generative Adversarial Networks make use of a unimodal latent/input distribution transformed by a continuous generator. Consequently, the modeled distribution always has connected support which is cumbersome when learning a disconnected set of manifolds. We formalize this problem by establishing a "no free lunch" theorem for the disconnected manifold learning stating an upper-bound on the precision of the targeted distribution. This is done by building on the necessary existence of a low-quality region where the generator continuously samples data between two disconnected modes. Finally, we derive a rejection sampling method based on the norm of generator’s Jacobian and show its efficiency on several generators including BigGAN.

Chat is not available.