Research map / Label-efficient and robust learning / Domain adaptation and generalization
Distribution-Matching Embedding for Visual Domain Adaptation
Mahsa Baktashmotlagh, Mehrtash Harandi, Mathieu Salzmann
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 nonlinear mapping (NL-DME-H instead of DME-H) | DME-H · Office+Caltech SURF kernel SVM | average recognition accuracy (%): 52.3 -> 52.5 | Improved |
| add nonlinear mapping (NL-DME-H instead of DME-H) | DME-H · Office+Caltech Decaf6 kernel SVM | average recognition accuracy (%): 86.7 -> 87.9 | Improved |
| use Hellinger distance instead of MMD (DME-H vs DME-MMD) | DME · Office+Caltech SURF kernel SVM | average recognition accuracy (%): 50.9 -> 52.3 | Improved |
| use Hellinger distance instead of MMD (DME-H vs DME-MMD) | DME · Office+Caltech Decaf7 linear SVM | average recognition accuracy (%): 87.2 -> 87.7 | Improved |
| replace Gaussian kernel with degree-2 polynomial kernel in MMD (ablation) | DME-MMD · Office+Caltech SURF kernel SVM | average recognition accuracy (%): 50.9 -> 50.4 | No clear effect |
| add distribution matching (DME-H) over no adaptation | NO ADAPT-SVM · Office+Caltech SURF kernel SVM | average recognition accuracy (%): 44 -> 52.3 | Improved |
| add distribution matching (NL-DME-H) over no adaptation | NO ADAPT-SVM · Office+Caltech Decaf8 kernel SVM | average recognition accuracy (%): 79.7 -> 82.4 | Improved |
| add distribution matching (NL-DME-H) over no adaptation | NO ADAPT-SVM · Office+Caltech Decaf8 linear SVM | average recognition accuracy (%): 77.6 -> 79.3 (SVMA 80.5 higher) | Hurt |
| use NL-DME-H instead of DDC deep domain confusion | DDC · Office+Caltech 6 pairs Decaf7 | average recognition accuracy (%): 84.6 -> 85.2 | Improved |
| use NL-DME-H instead of SA | SA · Office+Caltech Decaf7 kernel SVM | average recognition accuracy (%): 89.8 -> 88.3 | Hurt |
Closest papers by meaning
- Diversify and Match: A Domain Adaptive Representation Learning Paradigm for Object Detection (CVPR 2019 · similarity 0.89)
- Joint Geometrical and Statistical Alignment for Visual Domain Adaptation (CVPR 2017 · similarity 0.89)
- Learning an Invariant Hilbert Space for Domain Adaptation (CVPR 2017 · similarity 0.89)
- Preserving Clusters in Prompt Learning for Unsupervised Domain Adaptation (CVPR 2025 · similarity 0.88)
- A General Representation-Based Approach to Multi-Source Domain Adaptation (ICML 2025 · similarity 0.87)
- Adapting to Distribution Shift by Visual Domain Prompt Generation (ICLR 2024 · similarity 0.87)
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.
