Research map / Label-efficient and robust learning / Recognition and long-tail
On Learning Latent Models with Multi-Instance Weak Supervision
Kaifu Wang, Efthymia Tsamoura, Dan Roth
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 weak supervision (NLog) instead of fully supervised SL with same 4K samples | NLog · WEIGHTED-SUM M=2 | accuracy: 93 -> 97 | Improved |
| add pretraining with 16 directly labeled samples (ablation) | Scallop(1) · WEIGHTED-SUM M=3 | accuracy: 21 -> 97 | Improved |
| add pretraining with 16 directly labeled samples (ablation) | DLog-A · WEIGHTED-SUM M=2 | accuracy: 34 -> 80 | Improved |
| add pretraining with 16 directly labeled samples (ablation) | Scallop(4) · WEIGHTED-SUM M=4 | accuracy: 51 -> 78 | Improved |
| add pretraining with 8 or 16 labeled samples (ablation) | SL, DLog, NLog, NASP · WEIGHTED-SUM M=2 | accuracy: 97 -> 97 | No clear effect |
| increase number of digits M from 2 to 4 (no pretraining) (ablation) | Scallop(4) · WEIGHTED-SUM | accuracy: 97 -> 51 | Hurt |
| use top-k semantic loss with k=4 instead of k=1 (ablation) | Scallop · WEIGHTED-SUM M=4, #16 | accuracy: 54 -> 78 | Improved |
| use top-k=4 instead of k=1 without pretraining (ablation) | Scallop · WEIGHTED-SUM M=3, #0 | accuracy: 21 -> 53 | Improved |
| use DLog-A approximation instead of exact NLog (ablation) | DLog-A · WEIGHTED-SUM M=3, #0 | accuracy: 61 -> 24 | Hurt |
| use NLog top-5 approximation instead of exact semantic loss (ablation) | NLog(5) · WEIGHTED-SUM M=3, #0 | accuracy: 61 -> 27 | Hurt |
| use ABL or ENT instead of semantic-loss frameworks | ABL, ENT · WEIGHTED-SUM M=2 | accuracy: 97 -> 9 | Hurt |
| double SL training data from 4K to 8K | SL · WEIGHTED-SUM M=2 | accuracy: 93 -> 96 | Improved |
Closest papers by meaning
- Partial Label Causal Representation Learning for Instance-Dependent Supervision and Domain Generalization (AAAI 2025 · similarity 0.88)
- Learning with Partial-Label and Unlabeled Data: A Uniform Treatment for Supervision Redundancy and Insufficiency (ICML 2024 · similarity 0.87)
- Reduction-based Pseudo-label Generation for Instance-dependent Partial Label Learning (NeurIPS 2025 · similarity 0.87)
- Realistic Evaluation of Deep Partial-Label Learning Algorithms (ICLR 2025 · similarity 0.87)
- Distilling Reliable Knowledge for Instance-Dependent Partial Label Learning (AAAI 2024 · similarity 0.86)
- Conformal Prediction for Partial Label Learning (AAAI 2025 · similarity 0.86)
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.
