On the Geometry of Deep Learning

Balestriero, Randall, Humayun, Ahmed Imtiaz, Baraniuk, Richard

arXiv.org Artificial Intelligence 

Today's machine learning In this paper, we overview one promising avenue landscape is dominated by deep (neural) networks, of progress at the mathematical foundation of deep which are compositions of a large number of learning: the connection between deep networks and simple parameterized linear and nonlinear operators.