Research map / Label-efficient and robust learning / Semi-supervised learning
Auxiliary Task Reweighting for Minimum-data Learning
Baifeng Shi, Judy Hoffman, Kate Saenko, Trevor Darrell, Huijuan Xu
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 |
|---|---|---|---|
| Reweight Rotation and Exemplar-MT auxiliary tasks with ARML instead of uniform weights | S4L · CIFAR-10 (4000 labels) | test error: 15.67 ± .29 -> 13.68 ± .35 | Improved |
| Reweight Rotation and Exemplar-MT auxiliary tasks with ARML instead of uniform weights | S4L · SVHN (1000 labels) | test error: 7.83 ± .33 -> 5.89 ± .22 | Improved |
| Use ARML to reweight auxiliary tasks instead of the baseline weighting | CelebA 5_o_Clock_Shadow · CelebA (1% main-task labels) | test error: 6.70 ± .18 -> 5.97 ± .18 | Improved |
| Use ARML to reweight source domains without adding target labels to the classification loss | PACS domain generalization Baseline · PACS Sketch target (5 extra labels per class) | accuracy: 75.34 -> 79.35 | Improved |
| Add ARML reweighting of source domains | PACS joint training · PACS Art target (5 extra labels per class) | accuracy: 83.94 -> 85.70 | Improved |
| Add ARML reweighting of source domains | PACS joint training · PACS Cartoon target (5 extra labels per class) | accuracy: 81.36 -> 81.01 | Hurt |
| Add ARML reweighting of source domains | PACS joint training · PACS Photo target (5 extra labels per class) | accuracy: 97.01 -> 97.22 | Improved |
| Add ARML reweighting of source domains | PACS joint training · PACS Sketch target (5 extra labels per class) | accuracy: 78.52 -> 80.47 | Improved |
Closest papers by meaning
- Auxiliary Learning with Joint Task and Data Scheduling (ICML 2022 · similarity 0.87)
- AUXILIARY TASK UPDATE DECOMPOSITION: THE GOOD, THE BAD AND THE NEUTRAL (ICLR 2021 · similarity 0.85)
- Aux-NAS: Exploiting Auxiliary Labels with Negligibly Extra Inference Cost (ICLR 2024 · similarity 0.85)
- RECAST: Reparameterized, Compact weight Adaptation for Sequential Tasks (ICLR 2025 · similarity 0.83)
- Thumb on the Scale: Optimal Loss Weighting in Last Layer Retraining (NeurIPS 2025 · similarity 0.83)
- Optimizing importance weighting in the presence of sub-population shifts (ICLR 2025 · 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.
