1. we have an image of dimension 32*32*3. we want to collect 16*16*4 features from this image. we want to use a CONV layer of some 5*5 filters, then a POOL layer of 2*2 filter.

A) what will the stride, padding and filter dimensions of the CONV layer?

B) what will be the output dimension of the CONV layer?

C) what will be the stride, padding of the POOL layer?

D) how many parameters do we need for these two layers?

E) how many multiplications do we need for these layers?

2. Can we take advantage of transfer learning if the data set ( we are working on) is large? why or why not?