Research map / Label-efficient and robust learning / Domain adaptation and generalization
Sliced Wasserstein Discrepancy for Unsupervised Domain Adaptation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace MCD's L1 classifier discrepancy with sliced Wasserstein discrepancy | MCD-style domain adaptation · SVHN→MNIST | accuracy: 96.2 ± 0.4 -> 98.9 ± 0.1 | Helped |
| Replace MCD's L1 classifier discrepancy with sliced Wasserstein discrepancy | MCD-style domain adaptation · MNIST→USPS | accuracy: 94.4 ± 0.3 -> 98.6 ± 0.3 | Helped |
| Replace MCD's L1 classifier discrepancy with sliced Wasserstein discrepancy | MCD-style domain adaptation · USPS→MNIST | accuracy: 96.5 ± 0.3 -> 98.1 ± 0.1 | Helped |
| Replace MCD's L1 classifier discrepancy with sliced Wasserstein discrepancy | MCD-style domain adaptation · SYNSIG→GTSRB | accuracy: 94.1 ± 0.3 -> 97.1 ± 0.1 | Helped |
| Use task-specific classifier discrepancy alignment rather than source-only training | SWD · GTA5→Cityscapes | mIoU: 36.6 -> 44.5 | Helped |
| Replace MCD's discrepancy loss with sliced Wasserstein discrepancy | MCD-style object detection adaptation · VisDA 2018 object detection | mAP at 0.5 IoU: 4.7 -> 5.9 | Helped |
| Vary the number of radial projections M; authors report M=128 as sufficient for stable optimization and good accuracy (ablation) | SWD · SVHN→MNIST and SYNSIG→GTSRB | accuracy | No clear effect |
| Use SWD adaptation instead of source-only training | SWD · VisDA 2017 image classification | mean accuracy: 52.4 -> 76.4 | Helped |
| Use SWD adaptation instead of source-only training | SWD with VGG-16 · GTA5→Cityscapes | mIoU: 24.9 -> 39.9 | Helped |
Closest papers by meaning
- Unsupervised Domain Adaptation Using Feature-Whitening and Consensus Loss (CVPR 2019 · similarity 0.88)
- Maximum Classifier Discrepancy for Unsupervised Domain Adaptation (CVPR 2018 · similarity 0.88)
- Implicit Task-Driven Probability Discrepancy Measure for Unsupervised Domain Adaptation (NeurIPS 2021 · similarity 0.88)
- Unsupervised Domain Adaptation Based on Source-Guided Discrepancy (AAAI 2019 · similarity 0.87)
- Unsupervised Domain Adaptation With Distribution Matching Machines (AAAI 2018 · similarity 0.87)
- On $f$-Divergence Principled Domain Adaptation: An Improved Framework (NeurIPS 2024 · similarity 0.86)
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.