We have an input image of dimension 28X28X3. We want to use 5X5 filters to extract features. We need to extract 28X28X128 features.

A) What will be the dimension of the filters, stride, and padding?

B) How many multiplications do we need if we use just one CONV layer?

C) How can we reduce the number of multiplicatoins needed? Show how?

D) How much improvements can we get?

please answer as soon as possible