Problem $2(10)$

a$)$ If we have an input of size $1010,$ with $5$ channels, and the output is of dimensions $553,$ give two possible configurations of the weights and the layer. Can weights of dimension $44$ be used to get the output of the aforementioned dimensions? What about weights of dimension $55?(3)$

b$)$ Let us say we have an input of size $1213$ with $5$ channels. We first apply convolution with $6$ filters of dimension $46(1$ stride and no padding$).$ Then we apply convolution using $8$ filters of dimension $34(2$ stride and one padding$).$ Finally, we apply maxpooling using a filter of dimension $32.$ What is the final dimension of the output? $(7)$