Applies input filter$($s$)$ to the input image.

Args:

image: Tensor of shape $(1,$ d$1,$ h$1,$ w$1)$

kernel: Tensor of shape $($N$,$ d$1/$groups$,$ k$,$ k$)$ to be applied to the image

Returns:

filtered$\_$image: Tensor of shape $(1,$ d$2,$ h$2,$ w$2)$ where

d$2=$ N

h$2=($h$1-$ k $+2*$ padding$)/$ stride $+1$

w$2=($w$1-$ k $+2*$ padding$)/$ stride $+1$

HINTS:

$-$ You should use the $2$d convolution operator from torch.nn$.$functional.

$-$ In PyTorch, d$1$ is $`$in$\_$channels$`,$ and d$2$ is $`$out$\_$channels$`$

$-$ Make sure to pad the image appropriately $($it$\text{'}$s a parameter to the

convolution function you should use here!$).$

$-$ You can assume the number of groups is equal to the number of input channels.

$-$ You can assume only square filters for this function.