Research map / Label-efficient and robust learning / Domain adaptation and generalization
Domain Specified Optimization for Deployment Authorization
Haotian Wang, Haoang Chi, Wenjing Yang, Zhipeng Lin, Mingyang Geng, Long Lan, Jing Zhang, Dacheng Tao
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 |
|---|---|---|---|
| use DSO divergence-ball risk maximization instead of GNTL generative baseline | Source-only deployment authorization · Uniform authorization tasks (16 tasks) | successful authorization rate: 4/16 -> 15/16 | Improved |
| use DSO instead of GNTL on Office-31 trained on Amazon | Source-only authorization · Office-31 (Amazon) | T-Drop (%): 9.0% -> 30.0% | Improved |
| DSO versus GNTL on digit MNIST_M source | Source-only authorization · Digit (MNIST_M) | T-Drop (%): 35.7% -> 35.0% | Mixed |
| use TDSO with unlabeled unauthorized data instead of DSO | Target-combined authorization · Office-31 (Amazon) | T-Drop (%): 30.0% -> 67.5% | Improved |
| increase ratio of unlabeled target data for TDSO from 0.2 to 1.0 (ablation) | TDSO · Office-31 Dslr to Webcam | unauthorized accuracy: 36.9% -> 4.8% | Improved |
| random label corruption instead of Y_err = Y_S + 1 (ablation) | DSO · Office-31 and Visda2017 | authorization performance | No clear effect |
| vary radius rho from 50 to 300 (ablation) | DSO · VLCS, Digit, Visda2017 | authorization accuracy | No clear effect |
| apply DSO instead of purely supervised training | Source (authorized) domain performance · Six benchmarks | S-Drop: usually less than 3% | No clear effect |
Closest papers by meaning
- MAP: MAsk-Pruning for Source-Free Model Intellectual Property Protection (CVPR 2024 · similarity 0.83)
- A2XP: Towards Private Domain Generalization (CVPR 2024 · similarity 0.83)
- In Search of Forgotten Domain Generalization (ICLR 2025 · similarity 0.82)
- Jailbreaking Vision-Language Models via Dissonance-Guided Suffix Optimization and Image-Phrase Injection (CVPR 2026 · similarity 0.82)
- Multi-Source Collaborative Gradient Discrepancy Minimization for Federated Domain Generalization (AAAI 2024 · similarity 0.82)
- Vision-Language Model IP Protection via Prompt-based Learning (CVPR 2025 · similarity 0.82)
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.
