Research map / Vision-language and multimodal / Cross-modal learning
Asymmetric Cross-Modal Knowledge Distillation: Bridging Modalities with Weak Semantic Consistency
Riling Wei, Kelu Yao, Chuanguang Yang, Jin Wang, Zhuoyan Gao, Chao Li
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 self-supervised semantic matching (SSM) (ablation) | SemBridge · S2S-EU | OA (%): 92.5 -> 93.7 | Improved |
| add self-supervised semantic matching (SSM) (ablation) | SemBridge · S2S-CN | OA (%): 95.3 -> 96.2 | Improved |
| add dynamic matching (DynM) (ablation) | SemBridge · S2S-EU | OA (%): 92.9 -> 93.7 | Improved |
| add dynamic matching (DynM) (ablation) | SemBridge · M2S-GL | OA (%): 94.2 -> 96.6 | Improved |
| add L_ot1 alignment loss (ablation) | SemBridge · S2S-EU | OA (%): 92.5 -> 93.7 | Improved |
| add L_ot1 alignment loss (ablation) | SemBridge · S2S-CN | OA (%): 96.1 -> 96.2 | Improved |
| add L_ot2 alignment loss (ablation) | SemBridge · S2S-CN | OA (%): 94.1 -> 96.2 | Improved |
| add SemBridge to Vanilla KD | Vanilla KD ResNet34 · M2S-GL | OA (%): 93.6 -> 96.6 | Improved |
| add SemBridge to DKD | DKD ResNet34 · M2S-GL | OA (%): 68.1 -> 83.0 | Improved |
| add SemBridge to Logits | Logits ResNet34 · S2S-EU | OA (%): 87.8 -> 91.4 | Improved |
| add SemBridge to RKD | RKD ResNet34 · S2S-EU | OA (%): 91.6 -> 92.3 | Improved |
| use cross-modal distillation method VPR instead of uni-modal KD | VPR · S2S-EU | OA (%): 46.2 vs 92.6 (Vanilla KD) | Hurt |
Closest papers by meaning
- C2KD: Bridging the Modality Gap for Cross-Modal Knowledge Distillation (CVPR 2024 · similarity 0.90)
- Learning Modality Knowledge Alignment for Cross-Modality Transfer (ICML 2024 · similarity 0.87)
- DistilVPR: Cross-Modal Knowledge Distillation for Visual Place Recognition (AAAI 2024 · similarity 0.86)
- Align-KD: Distilling Cross-Modal Alignment Knowledge for Mobile Vision-Language Large Model Enhancement (CVPR 2025 · similarity 0.86)
- Beyond Modality Collapse: Representation Blending for Multimodal Dataset Distillation (NeurIPS 2025 · similarity 0.85)
- CovMatch: Cross-Covariance Guided Multimodal Dataset Distillation with Trainable Text Encoder (NeurIPS 2025 · similarity 0.85)
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.
