Research map / Vision-language and multimodal / Cross-modal learning
Unlearning in Cross-Modal Retrieval via Prior-Prototype Guided Partitioned Dampening
Yi Lu, Shu Li, Yurong Qian
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 uniform dampening (SSD) with prototype-guided hierarchical dampening (PPP) | SSD · COCO (dog), Dr | avg |delta| mAP vs Original: 2.07 -> 0.66 | Improved |
| replace uniform dampening (SSD) with PPP | SSD · Flickr25K (dog), Df | avg |delta| mAP vs Retrain: 10.8 -> 6.37 | Improved |
| replace uniform dampening (SSD) with PPP | SSD · DeepFashion (Trunks), Dr | accuracy (%): 79.5 -> 86.9 | Improved |
| replace SSD with PPP | SSD · ImageNet-100 (Doberman), Dr | accuracy (%): 96.4 -> 96.0 | No clear effect |
| increase number of forgotten classes from 1-2 to 4 and 8 (ablation) | PPP · COCO multi-class unlearning, Dr | mAP: 6.26% overall mAP drop with eight classes | Hurt |
| increase number of forgetting classes (ablation) | PPP · COCO multi-class unlearning, Df | mAP deviation from retrain | Mixed |
| increase forgetting sample size from 20 to 100 (ablation) | PPP · COCO, 64-bit T2I | confidence interval width on Df: 2.16 -> 0.56 | Improved |
| use only 20 forgetting samples (ablation) | PPP · COCO | mAP on Df and Dr | No clear effect |
Closest papers by meaning
- Label-Agnostic Forgetting: A Supervision-Free Unlearning in Deep Models (ICLR 2024 · similarity 0.87)
- Learning What to Forget: Improving LLM Unlearning via Learned Token-Level Importance (NeurIPS 2026 · similarity 0.87)
- Learning to Rematch Mismatched Pairs for Robust Cross-Modal Retrieval (CVPR 2024 · similarity 0.87)
- Fast Machine Unlearning without Retraining through Selective Synaptic Dampening (AAAI 2024 · similarity 0.87)
- Single Image Unlearning: Efficient Machine Unlearning in Multimodal Large Language Models (NeurIPS 2024 · similarity 0.87)
- Selective Forgetting: Advancing Machine Unlearning Techniques and Evaluation in Language Models (AAAI 2025 · similarity 0.87)
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.
