Gandalf: Learning label correlations in Extreme Multi-label Classification via Label Features

Siddhant Kharbanda, Devaansh Gupta, Erik Schultheis, Atmadeep Banerjee, Vikas Verma, Cho-Jui Hsieh, Rohit Babbar

Primary Area: general machine learning (i.e., none of the above)

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Keywords: Extreme Multilabel Classification, Key-phrase ads matching, short-text classification

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.

Abstract: Extreme Multi-label Text Classification (XMC) involves learning a classifier that can assign an input with a subset of most relevant labels from millions of label choices. Recent works in this domain have increasingly focused on a symmetric problem setting where both input instances and label features are short-text in nature. Short-text XMC with label features has found numerous applications in areas such as query-to-ad-phrase matching in search ads, title-based product recommendation, prediction of related searches, amongst others. In this paper, we propose Gandalf, a novel approach which makes use of a label correlation graph to leverage label features as additional data points to supplement the training distribution. By exploiting the characteristics of the short-text XMC problem, it leverages the label features to construct valid training instances, and uses the label graph for generating the corresponding soft-label targets, hence effectively capturing the label-label correlations. While most recent advances in XMC have been algorithmic, mainly aimed towards developing novel deep-learning frameworks, our data-centric augmentation approach is orthogonal to these methodologies, and can be applied in a plug-and-play manner to a variety of them. This generality and effectiveness of \textit{Gandalf} is demonstrated by showing up to 30\% relative improvements for 5 state-of-the-art algorithms across 4 benchmark datasets consisting of up to 1.3 million labels.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors' identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 9159