Correct $(1/1$ point$)$

Perceptron Update $2$

$0$ points possible $($ungraded$)$

For a given example $i,$ we defined the training error as $1$ if $y^{(i)}(*x^{(i)}+{}_0)0,$ and $0$ otherwise:

${}_i(,{}_0)=\left[\begin{array}{c}{y}^{(i)}(*x^{(i)}+{}_0)\end{array}\right]0$

Say we have a linear classifier given by $,{}_0.$ After the perceptron update using example $i,$ the training error ${}_i(,{}_0)$ for that example can $($select all those apply$)$:

Increase

Stay the same

Decrease

$$ You have used $0$ of $2$ attempts

Save