@inproceedings{luo-etal-2022-conditioned,
title = "Conditioned Masked Language and Image Modeling for Image-Text Dense Retrieval",
author = "Luo, Ziyang and
Xi, Yadong and
Zhang, Rongsheng and
Li, GongZheng and
Zhao, Zeng and
Ma, Jing",
editor = "Goldberg, Yoav and
Kozareva, Zornitsa and
Zhang, Yue",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2022",
month = dec,
year = "2022",
address = "Abu Dhabi, United Arab Emirates",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2022.findings-emnlp.10/",
doi = "10.18653/v1/2022.findings-emnlp.10",
pages = "130--140",
abstract = "Image-text retrieval is a fundamental cross-modal task that takes image/text as a query to retrieve relevant data of another type. The large-scale two-stream pre-trained models like CLIP have achieved tremendous success in this area. They embed the images and texts into instance representations with two separate encoders, aligning them on the instance-level with contrastive learning. Beyond this, the following works adopt the fine-grained token-level interaction (Masked Language and Image Modeling) to boost performance further. However, the vanilla token-level objectives are not designed to aggregate the image-text alignment information into the instance representations, but the token representations, causing a gap between pre-training and application. To address this issue, we carefully design two novel conditioned token-level pre-training objectives, Conditioned Masked Language and Image Modeling (ConMLM and ConMIM), forcing models to aggregate the token-level alignment information into the instance representations. Combing with the instance-level contrastive learning, we propose our cross-modal dense retrieval framework, Conditioned Language-Image Pre-training (ConLIP). Experimental results on two popular cross-modal retrieval benchmarks (MSCOCO and Flickr30k) reveal the effectiveness of our methods."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="luo-etal-2022-conditioned">
<titleInfo>
<title>Conditioned Masked Language and Image Modeling for Image-Text Dense Retrieval</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ziyang</namePart>
<namePart type="family">Luo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yadong</namePart>
<namePart type="family">Xi</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rongsheng</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">GongZheng</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zeng</namePart>
<namePart type="family">Zhao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jing</namePart>
<namePart type="family">Ma</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2022-12</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2022</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yoav</namePart>
<namePart type="family">Goldberg</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zornitsa</namePart>
<namePart type="family">Kozareva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Abu Dhabi, United Arab Emirates</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
</relatedItem>
<abstract>Image-text retrieval is a fundamental cross-modal task that takes image/text as a query to retrieve relevant data of another type. The large-scale two-stream pre-trained models like CLIP have achieved tremendous success in this area. They embed the images and texts into instance representations with two separate encoders, aligning them on the instance-level with contrastive learning. Beyond this, the following works adopt the fine-grained token-level interaction (Masked Language and Image Modeling) to boost performance further. However, the vanilla token-level objectives are not designed to aggregate the image-text alignment information into the instance representations, but the token representations, causing a gap between pre-training and application. To address this issue, we carefully design two novel conditioned token-level pre-training objectives, Conditioned Masked Language and Image Modeling (ConMLM and ConMIM), forcing models to aggregate the token-level alignment information into the instance representations. Combing with the instance-level contrastive learning, we propose our cross-modal dense retrieval framework, Conditioned Language-Image Pre-training (ConLIP). Experimental results on two popular cross-modal retrieval benchmarks (MSCOCO and Flickr30k) reveal the effectiveness of our methods.</abstract>
<identifier type="citekey">luo-etal-2022-conditioned</identifier>
<identifier type="doi">10.18653/v1/2022.findings-emnlp.10</identifier>
<location>
<url>https://aclanthology.org/2022.findings-emnlp.10/</url>
</location>
<part>
<date>2022-12</date>
<extent unit="page">
<start>130</start>
<end>140</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Conditioned Masked Language and Image Modeling for Image-Text Dense Retrieval
%A Luo, Ziyang
%A Xi, Yadong
%A Zhang, Rongsheng
%A Li, GongZheng
%A Zhao, Zeng
%A Ma, Jing
%Y Goldberg, Yoav
%Y Kozareva, Zornitsa
%Y Zhang, Yue
%S Findings of the Association for Computational Linguistics: EMNLP 2022
%D 2022
%8 December
%I Association for Computational Linguistics
%C Abu Dhabi, United Arab Emirates
%F luo-etal-2022-conditioned
%X Image-text retrieval is a fundamental cross-modal task that takes image/text as a query to retrieve relevant data of another type. The large-scale two-stream pre-trained models like CLIP have achieved tremendous success in this area. They embed the images and texts into instance representations with two separate encoders, aligning them on the instance-level with contrastive learning. Beyond this, the following works adopt the fine-grained token-level interaction (Masked Language and Image Modeling) to boost performance further. However, the vanilla token-level objectives are not designed to aggregate the image-text alignment information into the instance representations, but the token representations, causing a gap between pre-training and application. To address this issue, we carefully design two novel conditioned token-level pre-training objectives, Conditioned Masked Language and Image Modeling (ConMLM and ConMIM), forcing models to aggregate the token-level alignment information into the instance representations. Combing with the instance-level contrastive learning, we propose our cross-modal dense retrieval framework, Conditioned Language-Image Pre-training (ConLIP). Experimental results on two popular cross-modal retrieval benchmarks (MSCOCO and Flickr30k) reveal the effectiveness of our methods.
%R 10.18653/v1/2022.findings-emnlp.10
%U https://aclanthology.org/2022.findings-emnlp.10/
%U https://doi.org/10.18653/v1/2022.findings-emnlp.10
%P 130-140
Markdown (Informal)
[Conditioned Masked Language and Image Modeling for Image-Text Dense Retrieval](https://aclanthology.org/2022.findings-emnlp.10/) (Luo et al., Findings 2022)
ACL