Research map / Vision-language and multimodal / Cross-modal learning
A Graph Foundation Model with Cross-Modal Alignment and Modality-Aware Expert Fusion for Multi-Modal Graphs
Dongxiao He, Ankang Yang, Jitao Zhao, Di Jin
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 modality-specific GNN encoders instead of early fusion by mean pooling (ablation) | CAME · Amazon-sports | MRR: 26.83 -> 37.68 | Improved |
| add modality-specific GNN encoders instead of early fusion by mean pooling (ablation) | CAME · Amazon-cloth | MRR: 21.19 -> 28.88 | Improved |
| add dimension-wise gated fusion instead of average pooling (ablation) | CAME · Amazon-cloth | MRR: 27.90 -> 28.88 | Improved |
| add modality-aware MoE instead of parameter-matched MLP fusion (ablation) | CAME · Amazon-cloth | MRR: 26.37 -> 28.88 | Improved |
| add text-image alignment loss L_t,v (ablation) | CAME · Amazon-cloth | MRR: 20.64 -> 28.88 | Improved |
| add text-fusion alignment loss L_t,f (ablation) | CAME · Ele-fashion | accuracy: 85.04 -> 85.13 | No clear effect |
| use all three alignment losses instead of a single one (ablation) | CAME · Ele-fashion | accuracy: 80.74 -> 85.13 | Improved |
| use CAME instead of UniGraph2 (mean of CLIP features) | UniGraph2 · Amazon-sports | MRR: 29.52 -> 37.68 | Improved |
| use CAME on text-only graph | UniGraph2 · Wiki-CS 1-shot | accuracy: 35.56 -> 47.65 | Improved |
| use CAME instead of UniGraph2 | UniGraph2 · WN18RR | accuracy: 75.20 -> 81.97 | Improved |
| vary weights of L_t,f and L_v,f (ablation) | CAME · Ele-fashion and Amazon-cloth | performance | Mixed |
Closest papers by meaning
- Mario: Multimodal Graph Reasoning with Large Language Models (CVPR 2026 · similarity 0.91)
- Graph4MM: Weaving Multimodal Learning with Structural Information (ICML 2025 · similarity 0.91)
- Mosaic of Modalities: A Comprehensive Benchmark for Multimodal Graph Learning (CVPR 2025 · similarity 0.89)
- GRAPHGPT-O: Synergistic Multimodal Comprehension and Generation on Graphs (CVPR 2025 · similarity 0.88)
- GraphVLM: Benchmarking Vision Language Models for Multimodal Graph Learning (CVPR 2026 · similarity 0.88)
- Multi-View Empowered Structural Graph Wordification for 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.
