We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Moreau-Yosida $f$-divergences

Abstract: Variational representations of $f$-divergences are central to many machine learning algorithms, with Lipschitz constrained variants recently gaining attention. Inspired by this, we define the Moreau-Yosida approximation of $f$-divergences with respect to the Wasserstein-$1$ metric. The corresponding variational formulas provide a generalization of a number of recent results, novel special cases of interest and a relaxation of the hard Lipschitz constraint. Additionally, we prove that the so-called tight variational representation of $f$-divergences can be to be taken over the quotient space of Lipschitz functions, and give a characterization of functions achieving the supremum in the variational representation. On the practical side, we propose an algorithm to calculate the tight convex conjugate of $f$-divergences compatible with automatic differentiation frameworks. As an application of our results, we propose the Moreau-Yosida $f$-GAN, providing an implementation of the variational formulas for the Kullback-Leibler, reverse Kullback-Leibler, $\chi^2$, reverse $\chi^2$, squared Hellinger, Jensen-Shannon, Jeffreys, triangular discrimination and total variation divergences as GANs trained on CIFAR-10, leading to competitive results and a simple solution to the problem of uniqueness of the optimal critic.
Comments: ICML 2021 camera ready with appendix, 38 pages, 15 figures
Subjects: Machine Learning (cs.LG); Information Theory (cs.IT); Functional Analysis (math.FA); Optimization and Control (math.OC); Machine Learning (stat.ML)
MSC classes: 68T07 (Primary) 46N10, 94A17 (Secondary)
Cite as: arXiv:2102.13416 [cs.LG]
  (or arXiv:2102.13416v2 [cs.LG] for this version)

Submission history

From: Dávid Terjék [view email]
[v1] Fri, 26 Feb 2021 11:46:10 GMT (7373kb,D)
[v2] Fri, 11 Jun 2021 14:51:48 GMT (7223kb,D)

Link back to: arXiv, form interface, contact.