Choose all of the True statements regarding SGD.
O Using a small learning rate could cause the optimizer to converge more slowly.
O While optimizing a convex function using SGD, it is guaranteed that given enough time, the optimizer will always converge to the global minimum regardless of the learning rate.
O 1. Using the following sequence of learning rate will always result in converging to a local minimum.
a = 1/t +1
O The main idea behind SGD is that each step is always in the right direction when doing the approximation.