Consider training a single perceptron $($with sign activation function$)$ to linearly separate data points into two classes C$1($shown as $1)$ and C$2($shown as $-1).$ The samples $\{(0,1),(1,1)\}$ are associated with class C$1,$ and samples $\{(1,2),(2,2)\}$ are associate with class C$2.$ The initial weight vector is considered to be $(1,1)$ and the learning rate is $\backslash$eta $=1.$