Research map / Label-efficient and robust learning / Recognition and long-tail
Extreme F-measure Maximization using Sparse Probability Estimates
Kalina Jasinska, Krzysztof Dembczynski, Robert Busa-Fekete, Karlson Pfannschmidt, Timo Klerx, Eyke Hullermeier
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use OFO instead of STO threshold tuning (ablation) | PLT · RCV1 / Wiki10 / Amazon | macro F-score: 24.02 -> 30.28 | Improved |
| use OFO instead of STO threshold tuning (ablation) | PLT · Amazon | macro F-score: 46.94 -> 51.28 | Improved |
| use STO instead of FTA/OFO (ablation) | PLT · WikiLSHTC | macro F-score: 12.31 -> 16.22 | Improved |
| use FTA/OFO instead of STO (ablation) | PLT · AmazonCat | macro F-score: 31.64 -> 34.83 | Improved |
| replace FastXML with PLT under OFO | PLT · RCV1 | macro F-score: 18.93 -> 21.56 | Improved |
| replace FastXML with PLT under OFO | PLT · WikiLSHTC | macro F-score: 20.84 -> 14.00 | Hurt |
| replace FastXML with PLT under OFO | PLT · AmazonCat | macro F-score: 42.46 -> 33.13 | Hurt |
| replace FastXML with PLT | PLT · Wiki10 | Precision@1: 81.71 -> 84.34 | Improved |
| replace FastXML with PLT | PLT · Amazon | Precision@1: 34.24 -> 36.65 | Improved |
| replace FastXML with PLT | PLT · RCV1 | Precision@1: 91.13 -> 90.46 | Hurt |
| use STO with PLT instead of OFO on Wiki10 (ablation) | PLT · Wiki10 | number of inner products: 2518 -> 13682 | Hurt |
| use sparse probability estimates with OFO instead of FastXML | PLT · RCV1 | number of inner products: 747 -> 212 | Improved |
Closest papers by meaning
- Convex Calibrated Surrogates for the Multi-Label F-Measure (ICML 2020 · similarity 0.90)
- Revisiting F-measure Optimization in Multi-Label Classification: A Sampling-based Approach (CVPR 2026 · similarity 0.89)
- PD-Sparse : A Primal and Dual Sparse Approach to Extreme Multiclass and Multilabel Classification (ICML 2016 · similarity 0.88)
- Consistent algorithms for multi-label classification with macro-at-$k$ metrics (ICLR 2024 · similarity 0.85)
- Taming the Sigmoid Bottleneck: Provably Argmaxable Sparse Multi-Label Classification (AAAI 2024 · similarity 0.84)
- A General Online Algorithm for Optimizing Complex Performance Metrics (ICML 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
