Propensity-scored Probabilistic Label Trees
This work addresses the challenge of handling imbalanced and missing labels in web applications like content labeling and recommendation systems, representing an incremental improvement to existing XMLC methods.
The paper tackles the problem of making optimal predictions under the propensity model in extreme multi-label classification, specifically for probabilistic label trees, by introducing an inference procedure based on the A*-search algorithm that efficiently finds the optimal solution, and demonstrates its effectiveness in an empirical study on benchmark datasets.
Extreme multi-label classification (XMLC) refers to the task of tagging instances with small subsets of relevant labels coming from an extremely large set of all possible labels. Recently, XMLC has been widely applied to diverse web applications such as automatic content labeling, online advertising, or recommendation systems. In such environments, label distribution is often highly imbalanced, consisting mostly of very rare tail labels, and relevant labels can be missing. As a remedy to these problems, the propensity model has been introduced and applied within several XMLC algorithms. In this work, we focus on the problem of optimal predictions under this model for probabilistic label trees, a popular approach for XMLC problems. We introduce an inference procedure, based on the $A^*$-search algorithm, that efficiently finds the optimal solution, assuming that all probabilities and propensities are known. We demonstrate the attractiveness of this approach in a wide empirical study on popular XMLC benchmark datasets.