Hierarchy-aware Biased Bound Loss Function for Hierarchical Text Classification
Keywords: Deep Learning, Natural Language Processing, Hierarchical Text Classification
TL;DR: This paper presents the Hierarchy-aware Biased Bound (H2B) loss, a novel approach for unit-based hierarchical text classification models, addressing static thresholding and label imbalance.
Abstract: Hierarchical text classification (HTC) is a challenging problem with two key issues: utilizing structural information and mitigating label imbalance. Recently, the unit-based approach generating unit-based feature representations has outperformed the global approach focusing on a global feature representation. Nevertheless, unit-based models using BCE and ZMLCE losses still face static thresholding and label imbalance challenges. Those challenges become more critical in large-scale hierarchies. This paper introduces a novel hierarchy-aware loss function for unit-based HTC models: Hierarchy-aware Biased Bound (H2B) loss. H2B integrates learnable bounds and biases to address static thresholding and mitigate label imbalance adaptively. Experimental results on benchmark datasets demonstrate the superior performance of H2B compared to competitive HTC models.
Primary Area: representation learning for computer vision, audio, language, and other modalities
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.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2024/AuthorGuide.
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: 2255
Loading