Linear two-class (or multi-class) is another supervised learning technique which is similar to Linear Regression. The difference is mainly in the output of the dataset which is available as a two-class - binary (or multi-class) discrete values. Many applications fall in this category including face or object detection, sentiment analysis (positive or negative opinions), automatic diagnosis of medical conditions (presence or absence of a specific malady), etc. Also, non-linear classification can be used to separate instances that are not linearly separable.