Research map / Vision-language and multimodal / Cross-modal learning
Entity-Graph Enhanced Cross-Modal Pretraining for Instance-level Product Retrieval
Xiao ming Dong, Xunlin Zhan, Yunchao Wei, Xiao-Yong Wei, Yaowei Wang, Minlong Lu, Xiaochun Cao, Xiaodan Liang
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 masked entity modeling (E-Masked), image-text features (ablation) | EGE-CMP · Product1M multi-product retrieval | mAP: 81.49 -> 85.71 | Improved |
| add masked entity modeling (E-Masked), text features (ablation) | EGE-CMP · Product1M multi-product retrieval | mAP: 78.23 -> 83.48 | Improved |
| add masked entity modeling (E-Masked), image features (ablation) | EGE-CMP · Product1M multi-product retrieval | mAP: 80.94 -> 83.28 | Improved |
| add entity node loss (ablation) | EGE-CMP · Product1M multi-product retrieval | mAP: 85.71 -> 85.76 | No clear effect |
| add entity subgraph loss (ablation) | EGE-CMP · Product1M multi-product retrieval | mAP: 85.76 -> 86.91 | Improved |
| add entity subgraph loss, text features (ablation) | EGE-CMP · Product1M multi-product retrieval | mAP: 83.50 -> 85.25 | Improved |
| add single-modal transformer layers (0,6,6 -> 4,4,4) (ablation) | EGE-CMP · Product1M multi-product retrieval | mAP@10: 50.0 -> 86.9 | Improved |
| add cross-transformer layers (6,0,6 -> 4,4,4) (ablation) | EGE-CMP · Product1M multi-product retrieval | mAP@10: 85.9 -> 86.9 | Improved |
| add co-transformer layers (6,6,0 -> 4,4,4) (ablation) | EGE-CMP · Product1M multi-product retrieval | mAP@10: 85.8 -> 86.9 | Improved |
| scale to 24 layers (8,8,8) from 12 (ablation) | EGE-CMP · Product1M multi-product retrieval | mAP@10: 86.9 -> 86.9 | No clear effect |
| shrink to 6 layers (2,2,2) from 12 (ablation) | EGE-CMP · Product1M multi-product retrieval | mAP@10: 86.9 -> 83.7 | Hurt |
| replace UNITER with EGE-CMP on unseen brands | EGE-CMP · Product1M zero-shot (20 brands) | mAP@10: 43.3 -> 46.8 | Improved |
Closest papers by meaning
- Multiple Heads are Better than One: Mixture of Modality Knowledge Experts for Entity Representation Learning (ICLR 2025 · similarity 0.86)
- Tokenization, Fusion, and Augmentation: Towards Fine-grained Multi-modal Entity Representation (AAAI 2025 · similarity 0.85)
- Beyond Global Similarity: Multi-Conditional Retrieval for Fine-Grained Cross-Modal Understanding (CVPR 2026 · similarity 0.85)
- Heterogeneous Prompt-Guided Entity Inferring and Distilling for Scene-Text Aware Cross-Modal Retrieval (AAAI 2025 · similarity 0.85)
- Beyond Entities: A Large-Scale Multi-Modal Knowledge Graph with Triplet Fact Grounding (AAAI 2024 · similarity 0.84)
- Generative Multi-Modal Knowledge Retrieval with Large Language Models (AAAI 2024 · similarity 0.84)
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.
