Optimal Rates for Generalization of Gradient Descent for Deep ReLU Classification | Read Paper on Bytez