Toward a theory of optimization for over-parameterized systems of non-linear equations: the lessons of deep learning

Open in new window