Linear Multiclass Classification

Linear multiclass classification is a specific kind of targeted algorithm philosophy in machine learning and the field of structured prediction that uses both linear and multiclass methods. A multiclass classification is used to classify more than two classes – in contrast to a binary classification.

A linear classification uses an object’s characteristics to classify it by basing a decision on the value of a linear combination of characteristics.

With that said, the linear multiclass classification would apply that linear combination model to more than two classes. The whole structure for classifying more than two classes changes a lot from the procedure for binary classes. For instance, binary classification can use a confusion matrix and a set of four observed outcomes to create conclusions, where multiclass classification is a lot more complex.

Linear multiclass classification can be useful in structured prediction, which applies frameworks to problems where output variables are mutually dependent or constrained.