Research map / Label-efficient and robust learning / Recognition and long-tail
Learning from Label Proportions via Proportional Value Classification
Tianhao Ma, Wei Wang, Ximing Li, Gang Niu, Masashi Sugiyama
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 |
|---|---|---|---|
| replace proportion matching (PM) with proportional value classification | PM · K-MNIST (Random Bag, bag size 128) | accuracy: 28.30 -> 96.36 | Improved |
| replace PM with LLP-PVC | PM · CIFAR-10 (Random Bag, bag size 128) | accuracy: 49.21 -> 76.01 | Improved |
| replace ROT with LLP-PVC | ROT · CIFAR-10 (Random Bag, bag size 128) | accuracy: 33.70 -> 76.01 | Improved |
| replace ROT with LLP-PVC | ROT · SVHN (Cluster Bag, bag size 128) | accuracy: 58.33 -> 93.89 | Improved |
| replace ROT with LLP-PVC | ROT · K-MNIST (alpha-First Bag, bag size 128) | accuracy: 54.59 -> 96.03 | Improved |
| replace ROT with LLP-PVC at small bags | ROT · F-MNIST (Random Bag, bag size 4) | accuracy: 91.43 -> 91.39 | No clear effect |
| increase bag size from 4 to 128 for LLP-PVC | LLP-PVC · K-MNIST (Random Bag) | accuracy: 96.55 -> 96.36 | No clear effect |
| increase bag size from 4 to 128 for LLP-PVC | LLP-PVC · CIFAR-10 (Random Bag) | accuracy: 84.34 -> 76.01 | Hurt |
| add flooding regularization to EasyLLP | EasyLLP · CIFAR-10 (Random Bag, bag size 128) | accuracy: 37.36 -> 47.54 | Improved |
| add flooding regularization to EasyLLP | EasyLLP · K-MNIST (Cluster Bag, bag size 4) | accuracy: 31.61 -> 60.95 | Improved |
| use FFT divide-and-conquer instead of dynamic programming for the loss (ablation) | LLP-PVC (DP) · runtime vs bag size | execution time | Improved |
| use LLP-PVC instead of Count Loss | Count Loss · CIFAR-10 (bag size 8) | execution time: 139.52 -> 3.01 | Improved |
Closest papers by meaning
- Rethinking Learning from Label Proportions via Moment Matching (NeurIPS 2026 · similarity 0.93)
- Learning from Label Proportions: Bootstrapping Supervised Learners via Belief Propagation (ICLR 2024 · similarity 0.91)
- Forming Auxiliary High-confident Instance-level Loss to Promote Learning from Label Proportions (CVPR 2025 · similarity 0.91)
- Nearly Optimal Sample Complexity for Learning with Label Proportions (ICML 2025 · similarity 0.88)
- Class-Attribute Priors: Adapting Optimization to Heterogeneity and Fairness Objective (AAAI 2024 · similarity 0.84)
- Estimating Noisy Class Posterior with Part-level Labels for Noisy Label Learning (CVPR 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.
