Research map / Label-efficient and robust learning / Recognition and long-tail
Robust Data Programming with Precision-guided Labeling Functions
Oishik Chatterjee, Ganesh Ramakrishnan, Sunita Sarawagi
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 |
|---|---|---|---|
| decouple per-class theta parameters (CAGE-C-G) instead of Snorkel shared parameters | Snorkel · Spouse | F1: 0.41 -> 0.48 | Improved |
| add quality-guide regularizer (data independent) (ablation) | CAGE-C-G · SMS | F1: 0.34 -> 0.45 | Improved |
| add quality-guide regularizer (ablation) | CAGE-C-G · CDR | F1: 0.69 -> 0.69 | No clear effect |
| add continuous labeling functions (ablation) | CAGE-C · Ionosphere | F1: 0.82 -> 0.97 | Improved |
| add continuous labeling functions (ablation) | CAGE-C · Iris | F1: 0.87 -> 0.87 | No clear effect |
| replace sign penalty with proposed quality-guide regularizer (ablation) | CAGE-C-G · Spouse | F1: 0.48 -> 0.50 | Improved |
| use data-driven constraint for quality guide (ablation) | CAGE-C-G · Spouse | F1: 0.48 -> 0.48 | No clear effect |
| use undirected globally normalized discrete potentials instead of directed Bayesian network (ablation) | CAGE · Spouse | F1: 0.15 -> 0.58 | Improved |
| undirected vs directed potentials (ablation) | CAGE · SMS | F1: 0.59 -> 0.54 | Hurt |
| use full CAGE vs Snorkel | CAGE · Ionosphere | F1: 0.70 -> 0.97 | Improved |
| use quality guides to stabilize training across epochs (ablation) | CAGE-C · Spouse, SMS, CDR | F1 stability | Improved |
Closest papers by meaning
- Data Programming: Creating Large Training Sets, Quickly (NeurIPS 2016 · similarity 0.91)
- Learning from weak labelers as constraints (ICLR 2025 · similarity 0.86)
- Learning Hyper Label Model for Programmatic Weak Supervision (ICLR 2023 · similarity 0.86)
- Generative Modeling Reinvents Supervised Learning: Label Repurposing with Predictive Consistency Learning (ICML 2025 · similarity 0.86)
- Probably Approximately Precision and Recall Learning (NeurIPS 2025 · similarity 0.85)
- A Label is Worth A Thousand Images in Dataset Distillation (NeurIPS 2024 · similarity 0.85)
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.
