Research map / Optimization / Optimizers
Learning With Privileged Tasks
Yuru Song, Zan Lou, Shan You, Erkun Yang, Fei Wang, Chen Qian, Changshui Zhang, Xiaogang Wang
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 Pareto-MTL with PTL | MTL on Multi-Fashion · Multi-Fashion | target accuracy: 82.75% -> 86.71% | Improved |
| replace Pareto-MTL with PTL | MTL on Multi-MNIST · Multi-MNIST | target accuracy: 91.92% -> 97.80% | Improved |
| add privileged tasks (PTL) vs training target tasks only (tMOOMTL) | MOOMTL · CelebA (40 tasks) | target error: 28.55% -> 11.94% | Improved |
| add privileged tasks (PTL) vs training target tasks only (tMOOMTL) | MOOMTL · CIFAR-100 | target error: 17.76% -> 16.72% | Improved |
| add privileged tasks (PTL) vs training target tasks only (tMOOMTL) | MOOMTL · PASCAL | target accuracy: 81.30% -> 83.56% | Improved |
| add privileged tasks (PTL) vs training target tasks only (tMOOMTL) | MOOMTL · ImageNet-100 | target accuracy: 78.31% -> 79.46% | Improved |
| add privileged tasks (PTL) vs training target tasks only (tMOOMTL) | MOOMTL · Cityscapes | target accuracy: 64.35% -> 66.75% | Improved |
| replace MOOMTL with PTL | MOOMTL · Multi-FashionMNIST | target accuracy: 92.26% -> 96.09% | Improved |
| add slack descent on privileged tasks (tune C1 up to about 0.17) (ablation) | PTL, C2=0 · CIFAR-100 | target error: 16.52% | Improved |
| add direction correction toward target tasks (C2 about 0.82) (ablation) | PTL, C1=0.1 · CIFAR-100 | target error: 16.18% | Improved |
| replace STL with PTL | single-task learning · Multi-MNIST | target accuracy: 97.23% -> 97.80% | Improved |
Closest papers by meaning
- Quantifying Task Priority for Multi-Task Optimization (CVPR 2024 · similarity 0.87)
- Selective Task Group Updates for Multi-Task Optimization (ICLR 2025 · similarity 0.86)
- Fair Resource Allocation in Multi-Task Learning (ICML 2024 · similarity 0.85)
- Towards Consistent Multi-Task Learning: Unlocking the Potential of Task-Specific Parameters (CVPR 2025 · similarity 0.85)
- Delve into the Applicability of Advanced Optimizers for Multi-Task Learning (NeurIPS 2026 · similarity 0.85)
- Revisiting Fairness in Multitask Learning: A Performance-Driven Approach for Variance Reduction (CVPR 2025 · 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.
