$(1\text{'})$ Recall that for binary classification problems, the unweighted ensemble classifier of N

base classifier is defined as

h$\_($e$)($x$)=$argmax$\_($yin$\{-1,1\})(1)/($N$)\backslash$sum$\_($i$=1)^$N $1($h$\_($i$)($x$)=$y$).$

Table $1$ shows predictions by $3$ base classifiers h$\_(1),$h$\_(2),$h$\_(3)$ in $3$ different cases.

$($a$)(0\mathrm{.}5\text{'})$ Find the ensemble predictions $($shown as $"?"$ in the table$)$ and calculate the

accuracy of h$\_($e$)$ in all cases.

$($b$)(0\mathrm{.}5\text{'})$ Does the ensemble classifier always work better than the base classifier? What are

the conditions for an ensemble classifier to work better than the base classifier?

Case III

Table $1$: Ensemble classifier in three different cases.