1 paper · 2 filters
El Mehdi Achour, François Malgouyres, Sébastien Gerchinovitz
We study the optimization landscape of deep linear neural networks with the square loss. It is known that, under weak assumptions, there are no spurious local minima and no local m…