A Simple but Effective Closed-form Solution for Extreme Multi-label Learning
Kazuma Onishi, Katsuhiko Hayashi · 2025
Read A Simple but Effective Closed-form Solution for Extreme Multi-label Learning by Kazuma Onishi, Katsuhiko Hayashi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-10179.txt.
Description
Kazuma Onishi and Katsuhiko Hayashi offer a strikingly simple method for extreme multi label learning in a 2025 paper. The task assigns each item several labels from an enormous label set. Leading models carry many knobs to tune and are painful to reimplement. Their method rests on ridge regression, a classic technique with a closed form answer and a single tuning parameter.
A frequency based weighting, added in a few lines, lifts predictions for rare labels, which carry rich information but little data. On standard benchmarks, the simple method matched or beat heavily tuned rivals. The weighting clearly improved rare label accuracy. The authors share the code and present the result as a reminder that simple baselines deserve respect in this field.
Catalog record
| CREATOR | Kazuma Onishi, Katsuhiko Hayashi |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 18 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-10179