Saliency, LRP$,$ and DeepLIFT $(30$ points$).$

For all parts, use the neural network shown below.

Assume $i_1=-1,i_2=-2.$

$($a$)$ Compute the saliency values for $i_1$ and $i_2.$

$($b$)$ Compute the LRP feature relevance values for $i_1$ and $i_2($state the LRP rule you are using$).$

$($c$)$ Compute the DeepLIFT feature importance values for $i_1$ and $i_2($assuming the base input

of $i_1=i_2=0).$

Assume $i_1=1,i_2=2.$

$($d$)$ Compute the saliency values for $i_1$ and $i_2.$

$($e$)$ Compute the LRP feature relevance values for $i_1$ and $i_2($state the LRP rule you are using$).$

$($f$)$ Compute the DeepLIFT feature importance values for $i_1$ and $i_2($assuming the base input

of $i_1=i_2=0).$