Research map / Label-efficient and robust learning / Domain adaptation and generalization
GCA: Geometry-aware Conditional Alignment for Partial Domain Adaptation with Coding Rate Reduction
Xiaohui Chen, Chuan-Xian Ren
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 orthogonal discriminability loss LOrt (ablation) | GCA task loss · VisDA-2017 S→R | accuracy: 45.3 -> 92.0 | Improved |
| add conditional alignment loss LCA (ablation) | GCA task loss · VisDA-2017 S→R | accuracy: 45.3 -> 82.7 | Improved |
| add LCA to LOrt (full model) (ablation) | GCA · VisDA-2017 S→R | accuracy: 92.0 -> 93.5 | Improved |
| add LOrt to LCA (full model) (ablation) | GCA · Office-31 W→A | accuracy: 95.3 -> 96.4 | Improved |
| add LOrt and LCA (ablation) | GCA · ImageCLEF I→P | accuracy: 78.3 -> 88.1 | Improved |
| add LOrt only (ablation) | GCA task loss · ImageCLEF I→P | accuracy: 78.3 -> 80.3 | Improved |
| vary lambda1 and lambda2 in {2e-2..1e-1} (ablation) | GCA · Office-31 W→A | accuracy: avg 96.1, std 0.18, min 95.8, max 96.4 | No clear effect |
| use GCA instead of AR | AR · VisDA-2017 S→R | accuracy: 88.7 -> 93.5 | Improved |
| use GCA instead of AGAN | AGAN · Office-31 avg | accuracy: 97.2 -> 98.2 | Improved |
| use GCA instead of GATE/LEAD (geometry-based) | LEAD · Office-Home avg | accuracy: 73.8 -> 77.6 | Improved |
| use GCA on D→W and D→A | AGAN/ETN · Office-31 D→W | accuracy: 100 -> 99.9 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- LEAD — LEAD: Exploring Logit Space Evolution for Model Selection (CVPR 2024) · better on 4 of 4 shared results
- MLNet — MLNet: Mutual Learning Network with Neighborhood Invariance for Universal Domain Adaptation (AAAI 2024) · better on 1 of 1 shared result
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
- Theoretical Performance Guarantees for Partial Domain Adaptation via Partial Optimal Transport (ICML 2025 · similarity 0.87)
- Joint Geometrical and Statistical Alignment for Visual Domain Adaptation (CVPR 2017 · similarity 0.87)
- Gradient Harmonization in Unsupervised Domain Adaptation (TPAMI 2024 · similarity 0.87)
- CASUAL: Conditional Support Alignment for Domain Adaptation with Label Shift (AAAI 2025 · similarity 0.86)
- Invertible Projection and Conditional Alignment for Multi-Source Blended-Target Domain Adaptation (AAAI 2025 · similarity 0.86)
- Pairwise Alignment Improves Graph Domain Adaptation (ICML 2024 · similarity 0.85)
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.
