Research map / Label-efficient and robust learning / Recognition and long-tail
Achievement-Based Training Progress Balancing for Multi-Task Learning
Hayoung Yun, Hanjoo Cho
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 |
|---|---|---|---|
| add achievement-based task weight to DTP (ablation) | DTP · NYU v2 | AccMTL: 0.3660 -> 0.3745 | Improved |
| use weighted geometric mean of task losses instead of weighted sum (ablation) | DTP + achievement weight · NYU v2 | AccMTL: 0.3745 -> 0.3847 | Improved |
| add achievement-based weights to PCGrad (ablation) | PCGrad · NYU v2 | AccMTL: 0.3558 -> 0.3662 | Improved |
| add achievement-based weights to CAGrad (ablation) | CAGrad · NYU v2 | AccMTL: 0.3556 -> 0.3653 | Improved |
| replace arithmetic weighted sum with weighted geometric mean in DWA (ablation) | DWA · NYU v2 | AccMTL: 0.3676 -> 0.3811 | Improved |
| replace arithmetic weighted sum with weighted geometric mean in RLW (ablation) | RLW · NYU v2 | AccMTL: 0.3673 -> 0.3774 | Improved |
| replace arithmetic weighted sum with weighted geometric mean in DTP (ablation) | DTP · NYU v2 | AccMTL: 0.3660 -> 0.3800 | Improved |
| use gradient-based methods (MGDA, PCGrad, CAGrad) on partially annotated data | multi-task loss · VOC+NYU | AccMTL: 0.8642 -> 0.7621 (MGDA vs Uniform) | Hurt |
| use hard pseudo labels (MuST) instead of partially annotated labels | AMTL · VOC+NYU | AccMTL: 0.8784 -> 0.8431 | Hurt |
| add single-task teacher distillation (KD-MTL) to multi-task training | AMTL · VOC+NYU | AccMTL: 0.8784 -> 0.8742 | Hurt |
| apply any multi-task loss vs single-task baseline with surface normal task | multi-task losses · NYU v2 | AccMTL: 0.3989 -> 0.3847 (AMTL) | Hurt |
Closest papers by meaning
- Efficiently Identifying Task Groupings for Multi-Task Learning (NeurIPS 2021 · similarity 0.89)
- Towards Principled Task Grouping for Multi-Task Learning (NeurIPS 2026 · similarity 0.86)
- Revisiting Fairness in Multitask Learning: A Performance-Driven Approach for Variance Reduction (CVPR 2025 · similarity 0.85)
- Distribution Matching for Multi-Task Learning of Classification Tasks: A Large-Scale Study on Faces & Beyond (AAAI 2024 · similarity 0.85)
- AdaMerging: Adaptive Model Merging for Multi-Task Learning (ICLR 2024 · similarity 0.84)
- NTKMTL: Mitigating Task Imbalance in Multi-Task Learning from Neural Tangent Kernel Perspective (NeurIPS 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.
