Consider a multiclass classifier f that is constant as a function of the feature vector x$,$ that

is$,$ it always outputs the same label k in $\{1,...,$ K$\},$ where K is the number of classes. Let

$\backslash$pi k denote the prior class probability of class k$.$ In terms of $\backslash$pi $1,...,\backslash$pi K $,$ what class should the

classifier output to achieve the smallest probability of error $($amo