Research map / Bandits and online learning / Contextual bandits and Thompson sampling
Regret Bounds for Non-decomposable Metrics with Missing Labels
Nagarajan Natarajan, Prateek Jain
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 the thresholded max-likelihood estimator instead of the plugin estimator | Multi-label prediction · CAL500 | micro-F1: 0.4267 ± 0.0016 -> 0.3855 ± 0.0005 | Hurt |
| Use the thresholded max-likelihood estimator instead of the plugin estimator | Multi-label prediction · CAL500 | accuracy: 0.8541 ± 0.0034 -> 0.8493 ± 0.0002 | Hurt |
| Use the thresholded max-likelihood estimator instead of the plugin estimator | Multi-label prediction · Autofood | micro-F1: 0.4897 ± 0.0103 -> 0.5597 ± 0.0047 | Improved |
| Use the thresholded max-likelihood estimator instead of the plugin estimator | Multi-label prediction · Autofood | accuracy: 0.9307 ± 0.0064 -> 0.9345 ± 0.0043 | Improved |
| Use the thresholded max-likelihood estimator instead of the plugin estimator | Multi-label prediction · Bibtex | micro-F1: 0.2641 ± 0.0251 -> 0.2398 ± 0.0133 | Hurt |
| Use the thresholded max-likelihood estimator instead of the plugin estimator | Multi-label prediction · Bibtex | accuracy: 0.9849 ± 0.0016 -> 0.9856 ± 0.0003 | Improved |
| Use the thresholded max-likelihood estimator instead of the plugin estimator | Multi-label prediction · Compphys | micro-F1: 0.2463 ± 0.0315 -> 0.3510 ± 0.0293 | Improved |
| Use the thresholded max-likelihood estimator instead of the plugin estimator | Multi-label prediction · Compphys | accuracy: 0.9448 ± 0.0011 -> 0.9466 ± 0.0012 | Improved |
| Use the thresholded max-likelihood estimator instead of the plugin estimator | Multi-label prediction · Corel5k | micro-F1: 0.1552 ± 0.0116 -> 0.1642 ± 0.0001 | Improved |
| Use the thresholded max-likelihood estimator instead of the plugin estimator | Multi-label prediction · Corel5k | accuracy: 0.9906 ± 0.0000 -> 0.9906 ± 0.0000 | No clear effect |
Closest papers by meaning
- Regret Bounds for Multilabel Classification in Sparse Label Regimes (NeurIPS 2022 · similarity 0.88)
- Label Ranking through Nonparametric Regression (ICML 2022 · similarity 0.85)
- Convex Calibrated Surrogates for the Multi-Label F-Measure (ICML 2020 · similarity 0.85)
- Consistent algorithms for multi-label classification with macro-at-$k$ metrics (ICLR 2024 · similarity 0.84)
- Relaxing the Accurate Imputation Assumption in Doubly Robust Learning for Debiased Collaborative Filtering (ICML 2024 · similarity 0.84)
- Probably Approximately Precision and Recall Learning (NeurIPS 2025 · similarity 0.83)
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.
