Fetching the paper…

Towards moderate overparameterization: global convergence guarantees for training shallow neural networks · Around