Research map / Label-efficient and robust learning / Recognition and long-tail
Training Set Debugging Using Trusted Items
Xuezhou Zhang, Xiaojin Zhu, Stephen Wright
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 trusted items to jointly identify sparse training-label changes and retrain the learner | DUTI · Harry Potter toy classification | precision-recall | Improved |
| Use trusted items to jointly identify sparse training-label changes and retrain the learner | DUTI · German Loan | precision-recall | Improved |
| Use trusted items to jointly identify sparse training-label changes and retrain the learner | DUTI · Adult Income | precision-recall | Improved |
| Use trusted items to jointly identify sparse training-label changes and retrain the learner | DUTI · Handwritten Digits | bug detection precision-recall | Mixed |
| Use trusted items to jointly identify sparse training-label changes and retrain the learner | DUTI · Handwritten Digits | correct fixes | Improved |
| Use trusted items to jointly identify sparse training-label changes and retrain the learner | DUTI · Regression toy examples | precision-recall | Improved |
| Use the self-consistent training-loss term when proposing regression label fixes | DUTI · Regression toy examples | quality of suggested fixes | Improved |
| Replace DUTI's joint optimization with influence-function ranking | DUTI · Harry Potter toy classification | precision-recall | Hurt |
| Replace DUTI's joint optimization with nearest-neighbor heuristics | DUTI · Regression toy examples | bug detection precision-recall | Hurt |
Closest papers by meaning
- A Theory of Learning with Corrupted Labels (JMLR 2018 · similarity 0.82)
- Provably Reliable Conformal Prediction Sets in the Presence of Data Poisoning (ICLR 2025 · similarity 0.81)
- MIBP-Cert: Certified Training against Data Perturbations with Mixed-Integer Bilinear Programs (NeurIPS 2025 · similarity 0.80)
- Set Learning for Accurate and Calibrated Models (ICLR 2024 · similarity 0.80)
- Corruptions of Supervised Learning Problems: Typology and Mitigations (NeurIPS 2026 · similarity 0.80)
- SetAD: Semi-Supervised Anomaly Learning in Contextual Sets (NeurIPS 2026 · similarity 0.80)
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.
