Research map / Learning theory and algorithms / Prediction and decision losses
Mediated Uncoupled Learning: Learning Functions without Direct Input-output Correspondences
Ikko Yamane, Junya Honda, Florian Yger, 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 Naive with Joint-RR | Low-quality image classification · MNIST | accuracy: 88.06% -> 96.34% | Improved |
| Replace Naive with Joint-RR | Low-quality image classification · Fashion-MNIST | accuracy: 73.12% -> 86.93% | Improved |
| Replace Naive with Joint-RR | Low-quality image classification · CIFAR-10 | accuracy: 48.35% -> 69.10% | Improved |
| Replace Naive with Joint-RR | Low-quality image classification · CIFAR-100 | accuracy: 19.97% -> 28.05% | Improved |
| Replace Naive with Joint-RR | Low-quality image classification · MNIST | MSE: 0.184 -> 0.056 | Improved |
| Replace Naive with Joint-RR | Low-quality image classification · Fashion-MNIST | MSE: 0.417 -> 0.194 | Improved |
| Replace Naive with Joint-RR | Low-quality image classification · CIFAR-10 | MSE: 0.778 -> 0.424 | Improved |
| Replace Naive with Joint-RR | Low-quality image classification · CIFAR-100 | MSE: 0.935 -> 0.846 | Improved |
| Jointly train f and h instead of training them in two steps | 2Step-RR · MNIST | accuracy: 96.19% -> 96.34% | No clear effect |
| Jointly train f and h instead of training them in two steps | 2Step-RR · Fashion-MNIST | accuracy: 85.53% -> 86.93% | Improved |
| Jointly train f and h instead of training them in two steps | 2Step-RR · CIFAR-10 | accuracy: 67.60% -> 69.10% | Improved |
| Jointly train f and h instead of training them in two steps | 2Step-RR · CIFAR-100 | accuracy: 27.43% -> 28.05% | Improved |
Closest papers by meaning
- Uncoupled Regression from Pairwise Comparison Data (NeurIPS 2019 · similarity 0.85)
- Coupled Training with Privileged Information and Unlabeled Data (ICML 2026 · similarity 0.84)
- Inverse Entropic Optimal Transport Solves Semi-supervised Learning via Data Likelihood Maximization (ICML 2026 · similarity 0.84)
- Few-Shot Learning via Learning the Representation, Provably (ICLR 2021 · similarity 0.83)
- Learning Shared Representations from Unpaired Data (NeurIPS 2025 · similarity 0.82)
- Learnability of high-dimensional targets by two-parameter models and gradient flow (NeurIPS 2024 · similarity 0.82)
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.
