Hockey-Stick GANDownload PDF

12 Feb 2018, 18:57 (modified: 04 Jun 2018, 15:01)ICLR 2018 Workshop SubmissionReaders: Everyone
Keywords: generative models, GAN, hockey-stick divergence, information theory
TL;DR: New GAN objective with theoretical support from information theory.
Abstract: We propose a new objective for generative adversarial networks (GANs) that is aimed to address current issues in GANs such as mode collapse and unstable convergence. Our approach stems from the hockey-stick divergence that has properties we claim to be of great importance in generative models. We provide theoretical support for the model and preliminary results on synthetic Gaussian data.
4 Replies