One Vs All In K Classes

#Computers
For each class, use one hot encoding. The model chooses the class that has is predicted to be a 1, and if there's multiple such classes, randomly choose one.