Research map / Label-efficient and robust learning / Domain adaptation and generalization
Hierarchical Adaptive Structural SVM for Domain Adaptation
Jiaolong Xu, Sebastian Ramos, David Vázquez, Antonio Manuel López
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 hierarchical adaptation tree over target domains instead of pooling all targets (ablation) | A-SSVM-ALL · Office-Caltech (average over all splits) | accuracy: 54.4 -> 56.7 | Improved |
| pool all target domains instead of single target (ablation) | A-SSVM · Office-Caltech (average over all splits) | accuracy: 52.9 -> 54.4 | Improved |
| use three-layer adaptation tree A->[C,[D,W]] instead of two-layer (ablation) | HA-SSVM · Office-Caltech (A source) | average accuracy: 48.4 -> 49.5 | Improved |
| use three-layer adaptation tree A->[W,[D,C]] instead of two-layer (ablation) | HA-SSVM · Office-Caltech (A source) | average accuracy: 48.4 -> 47.7 | Hurt |
| use three-layer tree W->[D,[A,C]] instead of two-layer (ablation) | HA-SSVM · Office-Caltech (W source) | average accuracy: 47.3 -> 48.1 | Improved |
| use three-layer tree D->[W,[A,C]] instead of two-layer (ablation) | HA-SSVM · Office-Caltech (D source) | average accuracy: 48.7 -> 49.6 | Improved |
| use three-layer tree C->[A,[W,D]] instead of two-layer (ablation) | HA-SSVM · Office-Caltech (C source) | average accuracy: 57.3 -> 57.8 | Improved |
| use intermediate model N0 instead of A-SSVM (progressive adaptation) (ablation) | HA-SSVM-N0 · Caltech pedestrian detection | classification accuracy | Improved |
| replace given target domains with Reshape-Pr discovered domains (2 layers) (ablation) | HA-SSVM · Office-Caltech (C source) | accuracy: 57.3 -> 59.4 | Improved |
| replace given target domains with LatDD-Pr discovered domains (2 layers) (ablation) | HA-SSVM · Office-Caltech (A source) | accuracy: 48.4 -> 46.2 | Hurt |
| replace LatDD-Pr with Reshape-Pr for domain discovery (2 layers) (ablation) | HA-SSVM · Office-Caltech (A source) | accuracy: 46.2 -> 49.0 | Improved |
| use discovered domains Reshape-Pr with HA-SSVM instead of A-SSVM-ALL pooling (ablation) | HA-SSVM · Office-Caltech (A source) | accuracy: 47.6 -> 49.0 | Improved |
Closest papers by meaning
- Domain Generalization and Adaptation Using Low Rank Exemplar SVMs (TPAMI 2017 · similarity 0.88)
- Unsupervised Domain Adaptation of Object Detectors: A Survey (TPAMI 2023 · similarity 0.87)
- CAT: Exploiting Inter-Class Dynamics for Domain Adaptive Object Detection (CVPR 2024 · similarity 0.87)
- Scatter Component Analysis: A Unified Framework for Domain Adaptation and Domain Generalization (TPAMI 2016 · similarity 0.87)
- Universal Domain Adaptive Object Detection via Dual Probabilistic Alignment (AAAI 2025 · similarity 0.86)
- HiGDA: Hierarchical Graph of Nodes to Learn Local-to-Global Topology for Semi-Supervised Domain Adaptation (AAAI 2025 · 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.
