Research map / Label-efficient and robust learning / Domain adaptation and generalization
When Unsupervised Domain Adaptation Meets Tensor Representations
Hao Lu, Lei Zhang, Zhiguo Cao, Wei Wei, Ke Xian, Chunhua Shen, Anton van den Hengel
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 alignment matrices to naive tensor subspace learning (ablation) | NTSL · Office-Caltech10 mean | accuracy (%): 85.3 -> 86.9 | Improved |
| add alignment matrices to naive tensor subspace learning (ablation) | NTSL · ImageNet-VOC2007 mAP | average precision (%): 58.4 -> 58.7 | No clear effect |
| add alignment matrices to naive tensor subspace learning (ablation) | NTSL · Office-Caltech10 D->C | accuracy (%): 79.8 -> 84.0 | Improved |
| add alignment matrices to naive tensor subspace learning (ablation) | NTSL · Office-Caltech10 W->D | accuracy (%): 97.8 -> 97.7 | No clear effect |
| adapt tensor representation directly instead of vectorized subspace alignment (SA) | SA · Office-Caltech10 mean | accuracy (%): 74.3 -> 86.9 | Improved |
| adapt tensor representation directly instead of vectorized CORAL | CORAL · Office-Caltech10 mean | accuracy (%): 72.0 -> 86.9 | Improved |
| use TAISL instead of no adaptation on convolutional activations | NA · Office-Caltech10 mean | accuracy (%): 80.4 -> 86.9 | Improved |
| vector-form adaptation (PCA, TCA, DIP, SA, LTSL, CORAL) instead of no adaptation | NA · Office-Caltech10 mean | accuracy (%): 80.4 -> 41.4 | Hurt |
| use TAISL instead of LSSA on W->C | LSSA · Office-Caltech10 W->C | accuracy (%): 29.5 -> 82.6 | Improved |
| use TAISL instead of no adaptation on ImageNet-VOC2007 | NA · ImageNet-VOC2007 mAP | average precision (%): 51.6 -> 58.7 | Improved |
| use TAISL instead of LSSA on A->C | LSSA · Office-Caltech10 A->C | accuracy (%): 80.3 -> 80.1 | No clear effect |
| apply TAISL to HOG and VGG-M CONV5 tensor features (ablation) | baseline features · Office-Caltech10 (two tasks) | accuracy | Improved |
Closest papers by meaning
- Multivariate Regression on the Grassmannian for Predicting Novel Domains (CVPR 2016 · similarity 0.89)
- Incremental Multi-Domain Learning with Network Latent Tensor Factorization (AAAI 2020 · similarity 0.88)
- Learning an Invariant Hilbert Space for Domain Adaptation (CVPR 2017 · similarity 0.88)
- Representation Subspace Distance for Domain Adaptation Regression (ICML 2021 · similarity 0.88)
- LEARNING FACTORIZED REPRESENTATIONS FOR OPEN-SET DOMAIN ADAPTATION (ICLR 2019 · similarity 0.88)
- Domain Adaptation with Invariant Representation Learning: What Transformations to Learn? (NeurIPS 2021 · similarity 0.88)
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.
