Convex Calibrated Surrogates for Hierarchical Classification

Hierarchical classification problems are multiclass supervised learning problems with a pre-defined hierarchy over the set of class labels. In this work, we study the consistency of hierarchical class