Concept - One-vs-All Classification

(One-v.s.-All / One-v.s.-Rest, s.287)

• Multiclass problems have more than two classes, e.g. weather (cloudy, rain, snow), fruit (apple, orange, peach), email tagging (work, ad, friends) (s.285-286). One line cannot separate them (s.286).
• Method (s.287): train one Concept - Logistic Regression classifier $p_i(y=1|x)$ per class $i$ to predict the probability of $y=i$ (class $i$ vs all the rest); predict $i^{\ast }=\arg {\max }_i{p}_i(y=1|x)$.
• Alternative that models all classes jointly: Concept - Softmax Classifier.

Related

Concept - Softmax Classifier | Concept - Logistic Regression

Source chunk: 03 Text Classification - Chunk 10 (s.271-300)