Propensity-scored Probabilistic Label Trees

Marek Wydmuch, Kalina Jasinska-Kobus, Rohit Babbar, Krzysztof Dembczynski

Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

Abstract

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.
Original languageEnglish
Title of host publicationSIGIR '21: Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval
PublisherACM
Pages2252–2256
Number of pages5
ISBN (Electronic)978-1-4503-8037-9
DOIs
Publication statusPublished - 11 Jul 2021
MoE publication typeA4 Article in a conference publication
EventInternational ACM SIGIR Conference on Research and Development in Information Retrieval - Virtual, Online, Canada
Duration: 11 Jul 202115 Jul 2021
Conference number: 44

Conference

ConferenceInternational ACM SIGIR Conference on Research and Development in Information Retrieval
Abbreviated titleSIGIR
Country/TerritoryCanada
CityVirtual, Online
Period11/07/202115/07/2021

Fingerprint

Dive into the research topics of 'Propensity-scored Probabilistic Label Trees'. Together they form a unique fingerprint.

Cite this