Inducing a hierarchy for multi-class classification problems
Abstract
In applications where categorical labels follow a natural hierarchy, classification methods that exploit the label structure often outperform those that do not. Un-fortunately, the majority of classification datasets do not come pre-equipped with a hierarchical structure and classical flat classifiers must be employed. In this paper, we investigate a class of methods that induce a hierarchy that can similarly improve classification performance over flat classifiers. The class of methods follows the structure of first clustering the conditional distributions and subsequently using a hierarchical classifier with the induced hierarchy. We demonstrate the effectiveness of the class of methods both for discovering a latent hierarchy and for improving accuracy in principled simulation settings and three real data applications.
Cite
@article{arxiv.2102.10263,
title = {Inducing a hierarchy for multi-class classification problems},
author = {Hayden S. Helm and Weiwei Yang and Sujeeth Bharadwaj and Kate Lytvynets and Oriana Riva and Christopher White and Ali Geisa and Carey E. Priebe},
journal= {arXiv preprint arXiv:2102.10263},
year = {2021}
}