Research map / Vision-language and multimodal / Cross-modal learning
Multi-View Differential Mixing and Graph-Guided Structural Region Selection for Cross-Modal Alignment
Linlin Ji, Li Liu
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 average pooling with the multi-view differential mixer for global aggregation (ablation) | MG-Net · Flickr30K (Swin-Base-224) | rSum: 510.2 -> 544.0 | Improved |
| use max pooling instead of average pooling for global aggregation (ablation) | image-text matching · Flickr30K (Swin-Base-224) | rSum: 510.2 -> 541.7 | Improved |
| add differential views (C=2) on top of the base view only (ablation) | MDM · Flickr30K (Swin-Base-224) | rSum: 540.8 -> 544.0 | Improved |
| increase differential order from 2 to 3 (ablation) | MDM · Flickr30K (Swin-Base-224) | rSum: 544.0 -> 543.5 | Hurt |
| add the graph-guided region selector (GSRS) to the MDM (ablation) | MG-Net · Flickr30K (Swin-Base-224) | rSum: 544.0 -> 547.2 | Improved |
| add text guidance to the GSRS node features (ablation) | MG-Net · Flickr30K (Swin-Base-224) | rSum: 545.1 -> 547.2 | Improved |
| MG-Net versus PICO on Flickr30K with Swin-384 | MG-Net · Flickr30K (Swin-Base-384) | rSum: 548.2 -> 556.4 | Improved |
| MG-Net versus PICO on MS-COCO 1K with ViT-224 | MG-Net · MS-COCO 1K (ViT-Base-224) | rSum: 527.9 -> 528.1 | Improved |
| MG-Net versus PICO on MS-COCO 1K text-to-image retrieval | MG-Net · MS-COCO 1K (ViT-Base-224) | text-to-image R@5: 91.6 -> 90.8 | Hurt |
| MG-Net versus PICO on MS-COCO 5K image-to-text R@1 with Swin-224 | MG-Net · MS-COCO 5K (Swin-Base-224) | image-to-text R@1: 64.6 -> 67.9 | Improved |
Closest papers by meaning
- AlignMamba: Enhancing Multimodal Mamba with Local and Global Cross-modal Alignment (CVPR 2025 · similarity 0.87)
- CrossOver: 3D Scene Cross-Modal Alignment (CVPR 2025 · similarity 0.87)
- On the Value of Cross-Modal Misalignment in Multimodal Representation Learning (NeurIPS 2025 · similarity 0.87)
- Linguistic-Aware Patch Slimming Framework for Fine-grained Cross-Modal Alignment (CVPR 2024 · similarity 0.87)
- MM-Mixing: Multi-Modal Mixing Alignment for 3D Understanding (AAAI 2025 · similarity 0.87)
- CovMatch: Cross-Covariance Guided Multimodal Dataset Distillation with Trainable Text Encoder (NeurIPS 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.
