Problem $5\mathrm{.}2$ The loss $L$ for binary classification for a single training pair $\{x,y\}$ is:

$L=-(1-y)log[1-$sig$[f[x,]]]-ylog[sig[f[x,]]],$

where sig$[*]$ is defined in equation $5\mathrm{.}32.$ Plot this loss as a function of the transformed network output sig$[f[x,]]in[0,1]($i$)$ when the training label $y=0$ and $($ii$)$ when $y=1.$