Binary models can be adapted for multi$-$class classification. One$-$vs$-$Rest $($OvR$)$ is one way to adapt binary models for multi$-$class classification. OvR

sets more than one threshold for multi$-$class classification.

uses a variation of the logistic function to allow multi$-$class classification.

trains a binary model for each class to distinguish it from all other classes.

trains a binary model for each pair of classes.