Research map / Label-efficient and robust learning / Noisy labels
Improving Distantly Supervised Relation Extraction with Neural Noise Converter and Conditional Optimal Selector
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Add a learned neural noise converter (ablation) | PCNN with conditional optimal selector · NYT held-out test | Mean P@N: 78.8 -> 81.3 | Helped |
| Add a learned neural noise converter (ablation) | PCNN with conditional optimal selector · NYT held-out test | P@100: 81.0 -> 85.0 | Helped |
| Add a learned neural noise converter (ablation) | PCNN with conditional optimal selector · NYT held-out test | P@200: 78.5 -> 82.0 | Helped |
| Add a learned neural noise converter (ablation) | PCNN with conditional optimal selector · NYT held-out test | P@300: 77.0 -> 77.0 | No clear effect |
| Use conditional optimal selection instead of averaging positive-predicted sentence distributions (ablation) | PCNN with identity transition matrix · NYT held-out test | Mean P@N: 75.2 -> 78.8 | Helped |
| Use conditional optimal selection instead of averaging positive-predicted sentence distributions (ablation) | PCNN with identity transition matrix · NYT held-out test | P@300: 68.3 -> 77.0 | Helped |
| Use conditional optimal selection instead of averaging positive-predicted sentence distributions (ablation) | PCNN with identity transition matrix · NYT held-out test | P@100: 83.0 -> 81.0 | Hurt |
| Add conditional optimal selection to PCNN+ATT (ablation) | PCNN relation extraction · NYT held-out test | Mean P@N: 74.2 -> 78.8 | Helped |
| Use the full method instead of PCNN+ATT | PCNN relation extraction · NYT held-out test | Average precision: 36.5 -> 43.5 | Helped |
| Ensemble the 5 last saved models | PCNN+noise converter+conditional optimal selector · NYT held-out test | Average precision: 43.5 -> 45.2 | Helped |
Closest papers by meaning
- Are Noisy Sentences Useless for Distant Supervised Relation Extraction? (AAAI 2020 · similarity 0.91)
- Distantly Supervised Entity Relation Extraction with Adapted Manual Annotations (AAAI 2019 · similarity 0.89)
- Bootstrapping Distantly Supervised IE Using Joint Learning and Small Well-Structured Corpora (AAAI 2017 · similarity 0.88)
- Enhancing Relation Extraction via Supervised Rationale Verification and Feedback (AAAI 2025 · similarity 0.85)
- A Positive-Unlabeled Metric Learning Framework for Document-Level Relation Extraction with Incomplete Labeling (AAAI 2024 · similarity 0.85)
- Reliable Data Generation and Selection for Low-Resource Relation Extraction (AAAI 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.