Differentiable Programming A La Moreau

Vincent Roulet; Zaïd Harchaoui

ICASSP 2022accepted0 citations

Differentiable Programming A La Moreau

Abstract

The notion of a Moreau envelope is central to the analysis of first-order optimization algorithms for machine learning and signal processing. We define a compositional calculus adapted to Moreau envelopes and show how to apply it to deep networks, and, more broadly, to learning systems equipped with automatic differentiation and implemented in the spirit of differentiable programming.

BibTeX

@inproceedings{icassp2022_differentiablepr,
  title = {Differentiable Programming A La Moreau},
  author = {Vincent Roulet and Zaïd Harchaoui},
  booktitle = {ICASSP 2022},
  year = {2022}
}