Glossary term
Glossary term
Foundations
Given a classification problem with N classes, a solution consisting of N separate binary classification model—one binary classification model for each possible outcome. For example, given a model that classifies examples as animal, vegetable, or mineral, a one-vs.-all solution would provide the following three separate binary classification models:
animal versus not animal
vegetable versus not vegetable
mineral versus not mineral
Created for this library
An NLP team trains a one-vs-all classifier set for ticket routing where each binary classifier predicts membership in one team's queue.
A research team uses a one-vs-all decomposition on a moderate-size class problem as a baseline before evaluating multi-class models.
An image classification team uses one-vs-all classifiers for a small set of rare classes alongside a softmax head for common classes.
Definition source: Google for Developers Machine Learning Glossary | Creative Commons Attribution 4.0 License