Deep linear networks for regression are implicitly regularized towards flat minima

Devs

Deep linear networks for regression are implicitly regularized towards flat minima | Read Paper on Bytez