SGD Learns Over-parameterized Networks that Provably Generalize on Linearly Separable Data

Devs

SGD Learns Over-parameterized Networks that Provably Generalize on Linearly Separable Data | Read Paper on Bytez