Summary of Learning Label-label Correlations in Extreme Multi-label Classification Via Label Features, by Siddhant Kharbanda et al.
Learning label-label correlations in Extreme Multi-label Classification via Label Features
by Siddhant Kharbanda, Devaansh Gupta, Erik Schultheis, Atmadeep Banerjee, Cho-Jui Hsieh, Rohit Babbar
First submitted to arxiv on: 3 May 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Information Retrieval (cs.IR)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary This paper proposes Gandalf, a novel approach for Extreme Multi-label Text Classification (XMC), which leverages label features as additional data points to supplement the training distribution. The method uses a label co-occurrence graph to construct valid training instances and generate soft-label targets, capturing label-label correlations. Surprisingly, models trained on these new instances can outperform those trained on the original dataset, particularly on the PSP@k metric for tail labels. To further improve performance, the authors train six state-of-the-art algorithms on both the original and new training instances, achieving an average 5% relative improvement across four benchmark datasets containing up to 1.3M labels. Gandalf can be applied in a plug-and-play manner to various methods, advancing the state-of-the-art in XMC without additional computational overhead. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Gandalf is a new way to improve how computers classify text into many categories at once. This is important for things like recommending products based on search queries or matching ads with relevant phrases. The approach uses graphs to connect related labels and makes use of shorter text input instances. It’s surprising that models trained using this method can do better than those trained without it, especially when classifying less common labels. Overall, Gandalf is a step forward in making computers better at understanding many types of text. |
Keywords
» Artificial intelligence » Text classification
