A mathematical model for automatic differentiation in machine learning
Machine Learning
2020-10-30 v2 Optimization and Control
Machine Learning
Abstract
Automatic differentiation, as implemented today, does not have a simple mathematical model adapted to the needs of modern machine learning. In this work we articulate the relationships between differentiation of programs as implemented in practice and differentiation of nonsmooth functions. To this end we provide a simple class of functions, a nonsmooth calculus, and show how they apply to stochastic approximation methods. We also evidence the issue of artificial critical points created by algorithmic differentiation and show how usual methods avoid these points with probability one.
Cite
@article{arxiv.2006.02080,
title = {A mathematical model for automatic differentiation in machine learning},
author = {Jerome Bolte and Edouard Pauwels},
journal= {arXiv preprint arXiv:2006.02080},
year = {2020}
}