Pre-training Cross-Modal Retrieval by Expansive Lexicon-Patch Alignment

Yang Yiyuan, Guodong Long, Michael Blumenstein, Xiubo Geng, Chongyang Tao, Tao Shen, Daxin Jiang


Abstract
Recent large-scale vision-language pre-training depends on image-text global alignment by contrastive learning and is further boosted by fine-grained alignment in a weakly contrastive manner for cross-modal retrieval. Nonetheless, besides semantic matching learned by contrastive learning, cross-modal retrieval also largely relies on object matching between modalities. This necessitates fine-grained categorical discriminative learning, which however suffers from scarce data in full-supervised scenarios and information asymmetry in weakly-supervised scenarios when applied to cross-modal retrieval. To address these issues, we propose expansive lexicon-patch alignment (ELA) to align image patches with a vocabulary rather than only the words explicitly in the text for annotation-free alignment and information augmentation, thus enabling more effective fine-grained categorical discriminative learning for cross-modal retrieval. Experimental results show that ELA could effectively learn representative fine-grained information and outperform state-of-the-art methods on cross-modal retrieval.
Anthology ID:
2024.lrec-main.1136
Volume:
Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)
Month:
May
Year:
2024
Address:
Torino, Italia
Editors:
Nicoletta Calzolari, Min-Yen Kan, Veronique Hoste, Alessandro Lenci, Sakriani Sakti, Nianwen Xue
Venues:
LREC | COLING
SIG:
Publisher:
ELRA and ICCL
Note:
Pages:
12977–12987
Language:
URL:
https://aclanthology.org/2024.lrec-main.1136
DOI:
Bibkey:
Cite (ACL):
Yang Yiyuan, Guodong Long, Michael Blumenstein, Xiubo Geng, Chongyang Tao, Tao Shen, and Daxin Jiang. 2024. Pre-training Cross-Modal Retrieval by Expansive Lexicon-Patch Alignment. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024), pages 12977–12987, Torino, Italia. ELRA and ICCL.
Cite (Informal):
Pre-training Cross-Modal Retrieval by Expansive Lexicon-Patch Alignment (Yiyuan et al., LREC-COLING 2024)
Copy Citation:
PDF:
https://preview.aclanthology.org/proper-vol2-ingestion/2024.lrec-main.1136.pdf