Research map / Vision-language and multimodal / Human and embodied modeling
Graph Optimal Transport for Cross-Domain Alignment
Liqun Chen, Zhe Gan, Yu Cheng, Linjie Li, Lawrence Carin, Jingjing Liu
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 GOT to SCAN | SCAN image-text retrieval · Flickr30K and COCO | Rsum: 452.2 -> 474.8; 384.8 -> 403.9 | Improved |
| Combine WD and GWD in GOT rather than using either distance alone (ablation) | SCAN image-text retrieval · Flickr30K and COCO | Rsum: 472.3 -> 474.8; 402.6 -> 403.9 | Improved |
| Add GOT to BAN | BAN · VQA 2.0 validation | accuracy: 66.00 -> 66.44 | Improved |
| Add GOT to MLE | COCO image captioning · COCO | CIDEr: 106.3 -> 109.2 | Improved |
| Use GWD alone instead of the Transformer baseline (ablation) | Transformer machine translation · EN-VI uncased | BLEU: 29.25 ± 0.18 -> 28.65 ± 0.14 | Hurt |
| Add GOT to Transformer | Transformer machine translation · EN-VI uncased | BLEU: 29.25 ± 0.18 -> 29.92 ± 0.11 | Improved |
| Add GOT to Transformer | Transformer machine translation · EN-VI cased; EN-DE uncased and cased | BLEU: 28.46 ± 0.17 -> 29.09 ± 0.18; 25.60 ± 0.07 -> 26.05 ± 0.17; 25.12 ± 0.12 -> 25.54 ± 0.15 | Improved |
| Add GOT to LSTM | LSTM abstractive summarization · English Gigawords | ROUGE-1: 36.11 -> 37.10 | Improved |
| Add GOT to LSTM | LSTM abstractive summarization · English Gigawords | ROUGE-2: 16.39 -> 17.61 | Improved |
| Add GOT to LSTM | LSTM abstractive summarization · English Gigawords | ROUGE-L: 32.32 -> 33.70 | Improved |
| Use one shared transport plan for WD and GWD rather than separate plans (ablation) | GOT machine translation · EN-VI uncased and EN-DE uncased | BLEU: 29.77 ± 0.12 -> 29.92 ± 0.11; 25.89 ± 0.17 -> 26.05 ± 0.18 | Improved |
| Set λ to 0.8 instead of 0 (ablation) | GOT machine translation · EN-VI | BLEU: 28.65 -> 29.92 | Improved |
Closest papers by meaning
- A Unified Graph Language Model for Multi-Domain Multi-Task Graph Alignment Instruction Tuning (NeurIPS 2026 · similarity 0.88)
- Aligning Embeddings and Geometric Random Graphs: Informational Results and Computational Approaches for the Procrustes-Wasserstein Problem (NeurIPS 2024 · similarity 0.87)
- Pairwise Alignment Improves Graph Domain Adaptation (ICML 2024 · similarity 0.87)
- Iteratively Refined Early Interaction Alignment for Subgraph Matching based Graph Retrieval (NeurIPS 2024 · similarity 0.86)
- Hypergraph Learning for Unsupervised Graph Alignment via Optimal Transport (AAAI 2025 · similarity 0.86)
- Composing Object Relations and Attributes for Image-Text Matching (CVPR 2024 · similarity 0.86)
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.
