Deep linear neural networks with arbitrary loss: All local minima are global

Devs

Deep linear neural networks with arbitrary loss: All local minima are global | Read Paper on Bytez