Research map / Continual learning / Online and replay-based
Open-World Dynamic Prompt and Continual Visual Representation Learning
Youngeun Kim, Jun Fang, Qin Zhang, Zhaowei Cai, Yantao Shen, Rahul Duggal, Dripta S. Raychaudhuri, Zhuowen Tu, Yifan Xing, Onkar Dabeer
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 |
|---|---|---|---|
| replace static prompt pool with dynamic prompt generation network (frozen backbone) (ablation) | CodaPrompt · 100 CL stages avg | Recall@1: 67.95 -> 77.05 | Improved |
| add stage tokens to DPG (ablation) | DPG · 100 CL stages avg | Recall@1: 76.26 -> 77.05 | Improved |
| FIFO versus FILO or random replacement of stage tokens (ablation) | DPG · 100 CL stages avg | Recall@1: 77.05 vs 76.83 vs 77.01 | No clear effect |
| cap stage tokens at 5 instead of keeping all 100 (ablation) | DPG · 100 CL stages avg | Recall@1: 77.14 -> 77.05 | No clear effect |
| use low-rank (256) mapping function instead of full-rank (ablation) | DPG · In-Shop 10 stages | Recall@1: 71.40 -> 84.09 | Improved |
| fine-tune entire backbone instead of using DPG with frozen backbone (ablation) | DPaRL · 10 CL stages 4-dataset avg | Recall@1: 79.36 -> 69.58 | Hurt |
| update backbone with LoRA jointly with DPG (ablation) | DPG · 10 CL stages 4-dataset avg | Recall@1: 79.36 -> 80.80 | Improved |
| use LoRA rather than Adapter for backbone updating in DPaRL (ablation) | DPaRL · 10 CL stages 4-dataset avg | Recall@1: 78.70 -> 80.80 | Improved |
| DPaRL versus prior best prompt-based CL in 10 stages | CodaPrompt · 4 open-world retrieval benchmarks avg | Recall@1: 76.11 -> 80.80 | Improved |
| DPaRL versus prior prompt methods in closed-world medical domain | Coda · MedMNIST | top-1 accuracy: 54.49 -> 60.99 | Improved |
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.
- DualPrompt — DualPrompt: Complementary Prompting for Rehearsal-Free Continual Learning (ECCV 2022) · 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
- REP: Resource-Efficient Prompting for Rehearsal-Free Continual Learning (NeurIPS 2025 · similarity 0.86)
- OpenworldAUC: Towards Unified Evaluation and Optimization for Open-world Prompt Tuning (ICML 2025 · similarity 0.86)
- Convolutional Prompting meets Language Models for Continual Learning (CVPR 2024 · similarity 0.86)
- Evolving Parameterized Prompt Memory for Continual Learning (AAAI 2024 · similarity 0.86)
- Embracing Collaboration Over Competition: Condensing Multiple Prompts for Visual In-Context Learning (CVPR 2025 · similarity 0.86)
- CoPL: Contextual Prompt Learning for Vision-Language Understanding (AAAI 2024 · similarity 0.86)
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.
