Research map / Architectures and efficiency / CNN design
Many Task Learning With Task Routing
Gjorgji Strezoski, Nanne van Noord, Marcel Worring
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 partial sharing sigma=0.4 instead of full sharing sigma=1 (ablation) | Task Routing VGG-11 · UCSD-Birds (312 tasks) | accuracy: 59.6 -> 66.5 | Improved |
| add task routing sigma=0.2 instead of hard-shared sigma=1 (ablation) | Task Routing · Zappos50K (4 tasks) | accuracy: 79.2 -> 89.5 | Improved |
| add task routing sigma=0.2 instead of no sharing sigma=0 (ablation) | Task Routing · CelebA (10 tasks) | accuracy: 70.1 -> 73.2 | Improved |
| add partial sharing sigma=0.2 instead of sigma=0 (ablation) | Task Routing · CIFAR-10 (10 tasks) | accuracy: 96.5 -> 99.0 | Improved |
| use sigma=0.8 instead of 0.4 (ablation) | Task Routing · UCSD-Birds (50 tasks) | accuracy: 63.5 -> 57.2 | Hurt |
| use sigma=0.6 instead of 0.4 (ablation) | Task Routing · CelebA (40 tasks) | accuracy: 62.0 -> 65.3 | Improved |
| add task routing sigma=0.4 instead of modulation | Task Routing · UCSD-Birds (312 tasks) | accuracy: 56.7 -> 66.5 | Improved |
| add task routing sigma=0.2 instead of modulation | Task Routing · CIFAR-10 (10 tasks) | accuracy: 63.2 -> 99.0 | Improved |
| use cross-stitch (pretrained) instead of task routing | UCSD-Birds 10 tasks · UCSD-Birds (10 tasks) | accuracy: 65.1 -> 68.8 | Hurt |
| increase tasks from 10 to 40 | Task Routing · CelebA | accuracy: 73.2 -> 63.1 | Hurt |
| increase task count from 10 to 312 with sigma=0.4 | Task Routing · UCSD-Birds | accuracy: 65.1 -> 66.5 | Improved |
Closest papers by meaning
- MTFormer: Multi-task Learning via Transformer and Cross-Task Reasoning (ECCV 2022 · similarity 0.88)
- Routing Networks: Adaptive Selection of Non-Linear Functions for Multi-Task Learning (ICLR 2018 · similarity 0.87)
- TADFormer: Task-Adaptive Dynamic TransFormer for Efficient Multi-Task Learning (CVPR 2025 · similarity 0.86)
- Towards Principled Task Grouping for Multi-Task Learning (NeurIPS 2026 · similarity 0.86)
- Towards Consistent Multi-Task Learning: Unlocking the Potential of Task-Specific Parameters (CVPR 2025 · similarity 0.86)
- MTL-UE: Learning to Learn Nothing for Multi-Task Learning (ICML 2025 · similarity 0.85)
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.
