Timezone: »

 
Poster
A mathematical model for automatic differentiation in machine learning
Jérôme Bolte · Edouard Pauwels

Wed Dec 09 09:00 AM -- 11:00 AM (PST) @ Poster Session 3 #1029
in Poster Session 4 »

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.

Keywords: Information Theory · Learning Theory · Theory · Applications -> Privacy, Anonymity, and Security; Theory

Author Information

Jérôme Bolte (Université Toulouse Capitole and TSE)
Edouard Pauwels (IRIT)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors