# Joint Learning of Hyperbolic Label Embeddings for Hierarchical Multi-label Classification

Source: [Joint Learning of Hyperbolic Label Embeddings for Hierarchical Multi-label Classification](<https://www.sota2.com/research/paper/68fa690f1e209131cea1fdb5>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-01-04T03:23:33.522Z

Authors: Soumya Chatterjee, Ayush Maheshwari, Ganesh Ramakrishnan, Saketha Nath Jagaralpudi

Published: 2021-01-13T10:58:54.000Z

## Abstract

We consider the problem of multi-label classification where the labels lie in a hierarchy. However, unlike most existing works in hierarchical multi-label classification, we do not assume that the label-hierarchy is known. Encouraged by the recent success of hyperbolic embeddings in capturing hierarchical relations, we propose to jointly learn the classifier parameters as well as the label embeddings. Such a joint learning is expected to provide a twofold advantage: i) the classifier generalizes better as it leverages the prior knowledge of existence of a hierarchy over the labels, and ii) in addition to the label co-occurrence information, the label-embedding may benefit from the manifold structure of the input datapoints, leading to embeddings that are more faithful to the label hierarchy. We propose a novel formulation for the joint learning and empirically evaluate its efficacy. The results show that the joint learning improves over the baseline that employs label co-occurrence based pre-trained hyperbolic embeddings. Moreover, the proposed classifiers achieve state-of-the-art generalization on standard benchmarks. We also present evaluation of the hyperbolic embeddings obtained by joint learning and show that they represent the hierarchy more accurately than the other alternatives.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Hierarchical Text Classification on RCV1 (test)](<https://www.sota2.com/research/sota/hierarchical-text-classification-on-rcv1-test.md>): Hierarchical Text Classification; RCV1 (test); Micro-F1: 79.3.

- [Hierarchical Multi-label Classification on NYTimes](<https://www.sota2.com/research/sota/hierarchical-multi-label-classification-on-nytimes.md>): Hierarchical Multi-label Classification; NYTimes; Micro-F1: 77.

- [Hierarchical Multi-label Classification on Yelp](<https://www.sota2.com/research/sota/hierarchical-multi-label-classification-on-yelp.md>): Hierarchical Multi-label Classification; Yelp; Micro-F1: 60.8.

- [Hierarchical Text Classification on RCV1](<https://www.sota2.com/research/sota/hierarchical-text-classification-on-rcv1.md>): Hierarchical Text Classification; RCV1; Micro-F1: 79.3.

- [Hierarchical Text Classification on NYTimes (test)](<https://www.sota2.com/research/sota/hierarchical-text-classification-on-nytimes-test.md>): Hierarchical Text Classification; NYTimes (test); Micro-F1: 77.

- [Hierarchical Text Classification on Yelp (test)](<https://www.sota2.com/research/sota/hierarchical-text-classification-on-yelp-test.md>): Hierarchical Text Classification; Yelp (test); Micro-F1: 62.5.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2101.04997>)

[PDF](<https://arxiv.org/pdf/2101.04997.pdf>)

[Implementation](<https://github.com/soumyac1999/hyperbolic-label-emb-for-hmc>)
