Research map / Architectures and efficiency / Neural architecture search
MDL-NAS: A Joint Multi-Domain Learning Framework for Vision Transformer
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Search a joint task architecture rather than using the backbone searched for image classification (ablation) | MDL-NAS-B · COCO val | APb: 43.9 -> 44.6 | Helped |
| Search different architectures for tasks rather than one shared architecture (ablation) | MDL-NAS-B · COCO val | APb: 44.6 -> 45.0 | Helped |
| Add mask-based fine-grained parameter sharing instead of sharing all backbone parameters (ablation) | MDL-NAS-B · COCO val | APb: 43.9 -> 46.9 | Helped |
| Add sequential fine-grained parameter sharing instead of sharing all backbone parameters (ablation) | MDL-NAS-B · COCO val | APb: 43.9 -> 46.5 | Helped |
| Use fine-grained mask sharing instead of task-specific backbones (ablation) | MDL-NAS-B · ADE20K | mIoUss: 48.9 -> 49.9 | Helped |
| Freeze mask-sharing score parameters while fitting a new dataset (ablation) | MDL-NAS-B · CUB-200-2011 | Top-1 accuracy: 80.7 -> 86.8 | Helped |
| Vary mask-sharing score parameters while fitting a new dataset (ablation) | MDL-NAS-B · CUB-200-2011 | Top-1 accuracy: 80.7 -> 86.9 | Helped |
| Add sequential sharing when adapting to a new pose-estimation task (ablation) | MDL-NAS-B · Pose estimation | AP: 70.6 -> 74.0 | Helped |
Closest papers by meaning
- DATA: Domain-Aware and Task-Aware Self-Supervised Learning (CVPR 2022 · similarity 0.87)
- Learning Transferable Architectures for Scalable Image Recognition (CVPR 2018 · similarity 0.87)
- L-SWAG: Layer-Sample Wise Activation with Gradients Information for Zero-Shot NAS on Vision Transformers (CVPR 2025 · similarity 0.87)
- Learning Versatile Neural Architectures by Propagating Network Codes (ICLR 2022 · similarity 0.86)
- Transforming Vision Transformer: Towards Efficient Multi-Task Asynchronous Learner (NeurIPS 2024 · similarity 0.85)
- Win-Win: Training High-Resolution Vision Transformers from Two Windows (ICLR 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.