Research map / Label-efficient and robust learning / Domain adaptation and generalization
Disjoint Label Space Transfer Learning with Common Factorised Space
Xiaobin Chang, Yongxin Yang, Tao Xiang, Timothy M. Hospedales
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 |
|---|---|---|---|
| Use CFSM rather than Source only | SVHN→MNIST UDA · SVHN→MNIST | accuracy: 61.0 -> 86.5 | Improved |
| Use CFSM rather than Joint FT | SVHN→MNIST UDA · SVHN→MNIST | accuracy: 77.2 -> 86.5 | Improved |
| Use CFSM rather than LET | Semi-supervised DLSTL · SVHN (0–4)→MNIST (5–9), k=2 | accuracy: 91.7 ± 0.7 -> 93.5 ± 0.5 | Improved |
| Use CFSM rather than Source only | Sketchy SBIR · Sketchy target classes | Rank 1 accuracy: 23.74 ± 0.24 -> 26.22 ± 0.25 | Improved |
| Add the graph loss (ablation) | CFSM · Market→Duke (M2D) | Rank 1: 46.8 -> 49.8 | Improved |
| Replace conventional lower-level graph with the graph built on CFS (ablation) | CFSM · Market→Duke (M2D) | Rank 1: 47.4 -> 49.8 | Improved |
| Add target-domain data during regularised training (ablation) | CFSM · Market→Duke (M2D) | Rank 1: 41.6 -> 49.8 | Improved |
| Use factorisation and graph losses instead of feature reconstruction (ablation) | CFSM · Market→Duke (M2D) | Rank 1: 43.6 -> 49.8 | Improved |
| Add regularisers while training on source data only (ablation) | Source Only · Market→Duke (M2D) | Rank 1: 39.2 -> 41.6 | Improved |
Closest papers by meaning
- SLaM: Student-Label Mixing for Distillation with Unlabeled Examples (NeurIPS 2023 · similarity 0.85)
- Improve Unsupervised Pretraining for Few-Label Transfer (ICCV 2021 · similarity 0.85)
- Navigating Real-World Partial Label Learning: Unveiling Fine-Grained Images with Attributes (AAAI 2024 · similarity 0.84)
- Learning with Partial-Label and Unlabeled Data: A Uniform Treatment for Supervision Redundancy and Insufficiency (ICML 2024 · similarity 0.84)
- Disentangled Partial Label Learning (AAAI 2024 · similarity 0.83)
- Distilling Reliable Knowledge for Instance-Dependent Partial Label Learning (AAAI 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.
