Research map / Label-efficient and robust learning / Domain adaptation and generalization
Black-Box Unsupervised Domain Adaptation with Bi-Directional Atkinson-Shiffrin Memory
Jingyi Zhang, Jiaxing Huang, Xueying Jiang, Shijian Lu
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 sensory-short-term memory flows to conventional self-training (ablation) | black-box UDA self-training · GTA5 -> Cityscapes | mIoU: 35.2 -> 44.3 | Improved |
| add sensory-long-term memory flows to conventional self-training (ablation) | black-box UDA self-training · GTA5 -> Cityscapes | mIoU: 35.2 -> 44.5 | Improved |
| use both short-term and long-term memory with sensory memory (ablation) | BiMem · GTA5 -> Cityscapes | mIoU: 44.5 -> 46.4 | Improved |
| add forward memorization from short-term to long-term memory (ablation) | BiMem · GTA5 -> Cityscapes | mIoU: 46.4 -> 47.5 | Improved |
| add calibration of short-term memory by long-term memory (ablation) | BiMem · GTA5 -> Cityscapes | mIoU: 46.4 -> 47.4 | Improved |
| include all memorization and calibration flows (ablation) | BiMem · GTA5 -> Cityscapes | mIoU: 47.5 -> 48.2 | Improved |
| use BiMem versus DINE on classification, segmentation, detection | black-box UDA · Office-Home | Avg. accuracy: 69.7 -> 71.5 | Improved |
| use BiMem versus DINE on detection | black-box UDA · SYNTHIA -> Cityscapes detection | mAP: 29.1 -> 31.5 | Improved |
| vanilla self-training with noisy black-box pseudo labels | self-training · GTA5 -> Cityscapes | mIoU | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DINE — DINE: Domain Adaptation From Single and Multiple Black-Box Predictors (CVPR 2022) · better on 7 of 7 shared results
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
- Labeled Memory Networks for Online Model Adaptation (AAAI 2018 · similarity 0.86)
- MemFlow: A Lightweight Forward Memorizing Framework for Quick Domain Adaptive Feature Mapping (CVPR 2026 · similarity 0.86)
- ADU: Adaptive Detection of Unknown Categories in Black-Box Domain Adaptation (CVPR 2025 · similarity 0.85)
- A General Representation-Based Approach to Multi-Source Domain Adaptation (ICML 2025 · similarity 0.85)
- Black-Box Forgetting (NeurIPS 2024 · similarity 0.84)
- Preserving Clusters in Prompt Learning for Unsupervised Domain Adaptation (CVPR 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.
