Research map / Architectures and efficiency / RNNs and normalization
Learning to Branch for Multi-Task Learning
Pengsheng Guo, Chen-Yu Lee, Daniel Ulbricht
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 |
|---|---|---|---|
| Learn branching decisions instead of using the fixed Branch-VGG structure | VGG-19 multi-task model on CelebA · CelebA test | Acc (%): 90.79 -> 91.55 | Improved |
| Learn branching decisions instead of using the fixed GNAS-Deep-Wide structure | Deep-Wide multi-task model on CelebA · CelebA test | Acc (%): 91.36 -> 91.62 | Improved |
| Use learned branching rather than AdaShare's single-path sharing | LearnToBranch on Taskonomy · Taskonomy test | Segmentation: 0.566 -> 0.462 | Improved |
| Use learned branching rather than AdaShare's single-path sharing | LearnToBranch on Taskonomy · Taskonomy test | Normal: 0.707 -> 0.709 | Improved |
| Use learned branching rather than AdaShare's single-path sharing | LearnToBranch on Taskonomy · Taskonomy test | Depth: 0.025 -> 0.018 | Improved |
| Use learned branching rather than AdaShare's single-path sharing | LearnToBranch on Taskonomy · Taskonomy test | Keypoint: 0.192 -> 0.122 | Improved |
| Use learned branching rather than AdaShare's single-path sharing | LearnToBranch on Taskonomy · Taskonomy test | Edge: 0.193 -> 0.136 | Improved |
| Transfer the Taskonomy-learned grouping instead of learning grouping on NYUv2 | LearnToBranch on NYUv2 · NYUv2 test | Segmentation: 1.572 -> 1.611 | Hurt |
| Transfer the Taskonomy-learned grouping instead of learning grouping on NYUv2 | LearnToBranch on NYUv2 · NYUv2 test | Normal: 0.748 -> 0.739 | Hurt |
| Transfer the Taskonomy-learned grouping instead of learning grouping on NYUv2 | LearnToBranch on NYUv2 · NYUv2 test | Depth: 0.058 -> 0.058 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- AdaShare — AdaShare: Learning What To Share For Efficient Deep Multi-Task Learning (NeurIPS 2020) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- PackNet: Adding Multiple Tasks to a Single Network by Iterative Pruning (CVPR 2018 · similarity 0.85)
- Task Switching Network for Multi-Task Learning (ICCV 2021 · similarity 0.85)
- Latent Multi-Task Architecture Learning (AAAI 2019 · similarity 0.85)
- DMTG: One-Shot Differentiable Multi-Task Grouping (ICML 2024 · similarity 0.83)
- Selective Task Group Updates for Multi-Task Optimization (ICLR 2025 · similarity 0.83)
- Quantifying Task Priority for Multi-Task Optimization (CVPR 2024 · 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.
