Research map / Vision-language and multimodal / Cross-modal learning
On Modality Weighting and Specificity for Multi-Modal Entity Alignment
Yu Xing, Qizhuo Xie, Yunhui Liu, Qing Gu, Tao Zheng, Bin Chong, Tieke He
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 the graph structure modality (ablation) | HUMEA · FB15K-DB15K (20% seeds) | Hits@1: 0.4131 -> 0.5118 | Improved |
| add the visual modality (ablation) | HUMEA · FB15K-DB15K (20% seeds) | Hits@1: 0.4803 -> 0.5118 | Improved |
| add unimodal alignment loss (ablation) | HUMEA · FB15K-DB15K (20% seeds) | Hits@1: 0.4840 -> 0.5118 | Improved |
| add unimodal distillation loss (ablation) | HUMEA · FB15K-DB15K (20% seeds) | Hits@1: 0.4864 -> 0.5118 | Improved |
| add hierarchical MoE (ablation) | HUMEA · FB15K-DB15K (20% seeds) | Hits@1: 0.4778 -> 0.5118 | Improved |
| add CLUB expert disentanglement loss (ablation) | HUMEA · FB15K-DB15K (20% seeds) | Hits@1: 0.5033 -> 0.5118 | Improved |
| use bag-of-words relation/attribute features vs BERT text features (ablation) | HUMEA · FB15K-DB15K (20% seeds) | Hits@1: w/o rb 0.4736 vs w/o rt 0.4841 | Improved |
| set softmax temperature T away from about 0.1 (ablation) | HUMEA · FB15K-DB15K (20% seeds) | Hits@1 | Hurt |
| increase number of intra-MoE experts beyond a sufficient number (ablation) | HUMEA · FB15K-DB15K | Hits@1 | No clear effect |
| vary lambda1 and lambda2 (ablation) | HUMEA · FB15K-DB15K | Hits@1 | No clear effect |
Closest papers by meaning
- Tackling Uncertain Correspondences for Multi-Modal Entity Alignment (NeurIPS 2024 · similarity 0.95)
- Multiple Heads are Better than One: Mixture of Modality Knowledge Experts for Entity Representation Learning (ICLR 2025 · similarity 0.92)
- Pseudo-Label Calibration Semi-supervised Multi-Modal Entity Alignment (AAAI 2024 · similarity 0.90)
- Tokenization, Fusion, and Augmentation: Towards Fine-grained Multi-modal Entity Representation (AAAI 2025 · similarity 0.90)
- HLMEA: Unsupervised Entity Alignment Based on Hybrid Language Models (AAAI 2025 · similarity 0.89)
- MESED: A Multi-Modal Entity Set Expansion Dataset with Fine-Grained Semantic Classes and Hard Negative Entities (AAAI 2024 · similarity 0.88)
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.
