The reviewed record of science sign in
Pith

arxiv: 2008.10599 · v1 · pith:ISOHTS6M · submitted 2020-08-24 · cs.CV · cs.GR· cs.LG· cs.NE

The Hessian Penalty: A Weak Prior for Unsupervised Disentanglement

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:ISOHTS6Mrecord.jsonopen to challenge →

classification cs.CV cs.GRcs.LGcs.NE
keywords hessianpenaltyapplieddisentanglementlatenttermdeepencourages
0
0 comments X
read the original abstract

Existing disentanglement methods for deep generative models rely on hand-picked priors and complex encoder-based architectures. In this paper, we propose the Hessian Penalty, a simple regularization term that encourages the Hessian of a generative model with respect to its input to be diagonal. We introduce a model-agnostic, unbiased stochastic approximation of this term based on Hutchinson's estimator to compute it efficiently during training. Our method can be applied to a wide range of deep generators with just a few lines of code. We show that training with the Hessian Penalty often causes axis-aligned disentanglement to emerge in latent space when applied to ProGAN on several datasets. Additionally, we use our regularization term to identify interpretable directions in BigGAN's latent space in an unsupervised fashion. Finally, we provide empirical evidence that the Hessian Penalty encourages substantial shrinkage when applied to over-parameterized latent spaces.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.