Research map / Label-efficient and robust learning / Recognition and long-tail
Learning Multiple Tasks with Multilinear Relationship Networks
Mingsheng Long, ZHANGJIE CAO, Jianmin Wang, Philip S Yu
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 multilinear relationship learning in fc7 as well as fc8 (rather than only fc8) (ablation) | MRN · Office-Caltech 5%, 10%, and 20% train | average classification accuracy: 92.9 -> 93.8; 93.4 -> 94.5; 94.4 -> 95.5 | Improved |
| Add feature and class covariance modeling, not only task covariance (ablation) | MRN · Office-Caltech 5%, 10%, and 20% train | average classification accuracy: 92.7 -> 93.8; 93.4 -> 94.5; 93.6 -> 95.5 | Improved |
| Add multilinear relationship learning in fc7 as well as fc8 (rather than only fc8) (ablation) | MRN · Office-Home 5%, 10%, and 20% train | average classification accuracy: 56.3 -> 57.0; 62.4 -> 63.0; 66.7 -> 66.9 | Improved |
| Add feature and class covariance modeling, not only task covariance (ablation) | MRN · Office-Home 5%, 10%, and 20% train | average classification accuracy: 55.7 -> 57.0; 62.1 -> 63.0; 66.1 -> 66.9 | Improved |
| Add multilinear relationship learning in fc7 as well as fc8 (rather than only fc8) (ablation) | MRN · ImageCLEF-DA 5%, 10%, and 20% train | average classification accuracy: 67.7 -> 70.3; 71.2 -> 74.3; 73.7 -> 75.3 | Improved |
| Add feature and class covariance modeling, not only task covariance (ablation) | MRN · ImageCLEF-DA 5%, 10%, and 20% train | average classification accuracy: 69.1 -> 70.3; 73.3 -> 74.3; 74.0 -> 75.3 | Improved |
| Use tensor-normal multilinear relationships over task-specific layers | MRN (full) · Office-Caltech 5%, 10%, and 20% train | average classification accuracy: 89.4 -> 93.8; 92.0 -> 94.5; 93.3 -> 95.5 | Improved |
| Use tensor-normal multilinear relationships over task-specific layers | MRN (full) · Office-Home 5%, 10%, and 20% train | average classification accuracy: 53.4 -> 57.0; 60.8 -> 63.0; 66.5 -> 66.9 | Improved |
| Use tensor-normal multilinear relationships over task-specific layers | MRN (full) · ImageCLEF-DA 5%, 10%, and 20% train | average classification accuracy: 62.5 -> 70.3; 69.5 -> 74.3; 72.2 -> 75.3 | Improved |
Closest papers by meaning
- Deep Multi-task Representation Learning: A Tensor Factorisation Approach (ICLR 2017 · similarity 0.87)
- Provable Multi-Task Representation Learning by Two-Layer ReLU Neural Networks (ICML 2024 · similarity 0.85)
- Exploiting Task Relationships in Continual Learning via Transferability-Aware Task Embeddings (NeurIPS 2025 · similarity 0.85)
- Features are fate: a theory of transfer learning in high-dimensional regression (ICML 2025 · similarity 0.83)
- Quantifying Task Priority for Multi-Task Optimization (CVPR 2024 · similarity 0.83)
- Towards Principled Task Grouping for Multi-Task Learning (NeurIPS 2026 · similarity 0.83)
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.
