Research map / Label-efficient and robust learning / Domain adaptation and generalization
Adversarial Support Alignment
Read the paperCheck 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 |
|---|---|---|---|
| Use support alignment (ASA-abs) instead of relaxed distribution alignment (sDANN-4) | UDA · USPS→MNIST, α=2.0 | average class accuracy: 81.5 -> 90.4 | Helped |
| Use support alignment (ASA-abs) instead of relaxed distribution alignment (sDANN-4) | UDA · USPS→MNIST, α=2.0 | minimum class accuracy: 39.3 -> 68.4 | Helped |
| Use support alignment (ASA-sq) instead of relaxed distribution alignment (sDANN-4) | UDA · STL→CIFAR, α=2.0 | average class accuracy: 66.4 -> 68.1 | Helped |
| Use support alignment (ASA-sq) instead of relaxed distribution alignment (sDANN-4) | UDA · STL→CIFAR, α=2.0 | minimum class accuracy: 39.0 -> 44.7 | Helped |
| Use support alignment (ASA-abs) instead of relaxed distribution alignment (sDANN-4) | UDA · VisDA17, α=2.0 | average class accuracy: 50.7 -> 52.5 | Helped |
| Use support alignment (ASA-abs) instead of relaxed distribution alignment (sDANN-4) | UDA · VisDA17, α=2.0 | minimum class accuracy: 18.6 -> 19.7 | Helped |
| Use absolute rather than squared distance in ASA (ablation) | ASA · USPS→MNIST, α=2.0 | average class accuracy: 87.2 -> 90.4 | Helped |
| Use absolute rather than squared distance in ASA (ablation) | ASA · STL→CIFAR, α=2.0 | average class accuracy: 68.1 -> 67.8 | Hurt |
| Use absolute rather than squared distance in ASA (ablation) | ASA · VisDA17, α=2.0 | average class accuracy: 51.9 -> 52.5 | Helped |
| Use distribution alignment (DANN/VADA) rather than no adaptation | UDA · Positive label-shift levels across tasks | minimum class accuracy | Hurt |
Closest papers by meaning
- Domain Adaptation with Asymmetrically-Relaxed Distribution Alignment (ICML 2019 · similarity 0.90)
- Cooperative Distribution Alignment via JSD Upper Bound (NeurIPS 2022 · similarity 0.89)
- Cross-Domain Visual Representations via Unsupervised Graph Alignment (AAAI 2019 · similarity 0.87)
- CASUAL: Conditional Support Alignment for Domain Adaptation with Label Shift (AAAI 2025 · similarity 0.86)
- Log-Likelihood Ratio Minimizing Flows: Towards Robust and Quantifiable Neural Distribution Alignment (NeurIPS 2020 · similarity 0.86)
- Optimizing Distributional Geometry Alignment with Optimal Transport for Generative Dataset Distillation (NeurIPS 2025 · similarity 0.84)
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.