M3P: Learning Universal Representations via Multitask Multilingual Multimodal Pre-training

We present M3P, a Multitask Multilingual Multimodal Pre-trained model that combines multilingual pre-training and multimodal pre-training into a unified framework via multitask pre-training. Our goal is to learn universal representations that can map objects occurred in different modalities or texts...

Ausführliche Beschreibung

Gespeichert in:

Bibliographische Detailangaben
Hauptverfasser:	Ni, Minheng, Huang, Haoyang, Su, Lin, Cui, Edward, Bharti, Taroon, Wang, Lijuan, Gao, Jianfeng, Zhang, Dongdong, Duan, Nan
Format:	Artikel
Sprache:	eng
Schlagworte:	Computer Science - Computation and Language Computer Science - Computer Vision and Pattern Recognition
Online-Zugang:	Volltext bestellen
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

container_end_page
container_issue
container_start_page
container_title
container_volume
creator	Ni, Minheng Huang, Haoyang Su, Lin Cui, Edward Bharti, Taroon Wang, Lijuan Gao, Jianfeng Zhang, Dongdong Duan, Nan
description	We present M3P, a Multitask Multilingual Multimodal Pre-trained model that combines multilingual pre-training and multimodal pre-training into a unified framework via multitask pre-training. Our goal is to learn universal representations that can map objects occurred in different modalities or texts expressed in different languages into a common semantic space. In addition, to explicitly encourage fine-grained alignment between images and non-English languages, we also propose Multimodal Code-switched Training (MCT) to combine monolingual pre-training and multimodal pre-training via a code-switch strategy. Experiments are performed on the multilingual image retrieval task across two benchmark datasets, including MSCOCO and Multi30K. M3P can achieve comparable results for English and new state-of-the-art results for non-English languages.
doi_str_mv	10.48550/arxiv.2006.02635
format	Article
fullrecord	<record><control><sourceid>arxiv_GOX</sourceid><recordid>TN_cdi_arxiv_primary_2006_02635</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2006_02635</sourcerecordid><originalsourceid>FETCH-LOGICAL-a675-47ec9bbec30f213804e25047be7a5860c64e87ade692a78c377fa0dac2848aaa3</originalsourceid><addsrcrecordid>eNotj8tOwzAURL1hgQofwAr_QILrd9mhipeUigq1m26iG-cGWaRuZbsR_D1pympmpJmRDiF3c1ZKqxR7gPjjh5IzpkvGtVDXZLcS60daIcTgwxfdBj9gTNDTTzxGTBgyZH8IiQ4e6OrUZ58hfV9cPy5OY3UK-0M72nXEIkfw57MbctVBn_D2X2dk8_K8Wb4V1cfr-_KpKkAbVUiDbtE06ATr-FxYJpErJk2DBpTVzGmJ1kCLesHBWCeM6YC14LiVFgDEjNxfbie4-hj9HuJvfYasJ0jxBxfdTio</addsrcrecordid><sourcetype>Open Access Repository</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype></control><display><type>article</type><title>M3P: Learning Universal Representations via Multitask Multilingual Multimodal Pre-training</title><source>arXiv.org</source><creator>Ni, Minheng ; Huang, Haoyang ; Su, Lin ; Cui, Edward ; Bharti, Taroon ; Wang, Lijuan ; Gao, Jianfeng ; Zhang, Dongdong ; Duan, Nan</creator><creatorcontrib>Ni, Minheng ; Huang, Haoyang ; Su, Lin ; Cui, Edward ; Bharti, Taroon ; Wang, Lijuan ; Gao, Jianfeng ; Zhang, Dongdong ; Duan, Nan</creatorcontrib><description>We present M3P, a Multitask Multilingual Multimodal Pre-trained model that combines multilingual pre-training and multimodal pre-training into a unified framework via multitask pre-training. Our goal is to learn universal representations that can map objects occurred in different modalities or texts expressed in different languages into a common semantic space. In addition, to explicitly encourage fine-grained alignment between images and non-English languages, we also propose Multimodal Code-switched Training (MCT) to combine monolingual pre-training and multimodal pre-training via a code-switch strategy. Experiments are performed on the multilingual image retrieval task across two benchmark datasets, including MSCOCO and Multi30K. M3P can achieve comparable results for English and new state-of-the-art results for non-English languages.</description><identifier>DOI: 10.48550/arxiv.2006.02635</identifier><language>eng</language><subject>Computer Science - Computation and Language ; Computer Science - Computer Vision and Pattern Recognition</subject><creationdate>2020-06</creationdate><rights>http://arxiv.org/licenses/nonexclusive-distrib/1.0</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><link.rule.ids>228,230,780,885</link.rule.ids><linktorsrc>$$Uhttps://arxiv.org/abs/2006.02635$$EView_record_in_Cornell_University$$FView_record_in_$$GCornell_University$$Hfree_for_read</linktorsrc><backlink>$$Uhttps://doi.org/10.48550/arXiv.2006.02635$$DView paper in arXiv$$Hfree_for_read</backlink></links><search><creatorcontrib>Ni, Minheng</creatorcontrib><creatorcontrib>Huang, Haoyang</creatorcontrib><creatorcontrib>Su, Lin</creatorcontrib><creatorcontrib>Cui, Edward</creatorcontrib><creatorcontrib>Bharti, Taroon</creatorcontrib><creatorcontrib>Wang, Lijuan</creatorcontrib><creatorcontrib>Gao, Jianfeng</creatorcontrib><creatorcontrib>Zhang, Dongdong</creatorcontrib><creatorcontrib>Duan, Nan</creatorcontrib><title>M3P: Learning Universal Representations via Multitask Multilingual Multimodal Pre-training</title><description>We present M3P, a Multitask Multilingual Multimodal Pre-trained model that combines multilingual pre-training and multimodal pre-training into a unified framework via multitask pre-training. Our goal is to learn universal representations that can map objects occurred in different modalities or texts expressed in different languages into a common semantic space. In addition, to explicitly encourage fine-grained alignment between images and non-English languages, we also propose Multimodal Code-switched Training (MCT) to combine monolingual pre-training and multimodal pre-training via a code-switch strategy. Experiments are performed on the multilingual image retrieval task across two benchmark datasets, including MSCOCO and Multi30K. M3P can achieve comparable results for English and new state-of-the-art results for non-English languages.</description><subject>Computer Science - Computation and Language</subject><subject>Computer Science - Computer Vision and Pattern Recognition</subject><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2020</creationdate><recordtype>article</recordtype><sourceid>GOX</sourceid><recordid>eNotj8tOwzAURL1hgQofwAr_QILrd9mhipeUigq1m26iG-cGWaRuZbsR_D1pympmpJmRDiF3c1ZKqxR7gPjjh5IzpkvGtVDXZLcS60daIcTgwxfdBj9gTNDTTzxGTBgyZH8IiQ4e6OrUZ58hfV9cPy5OY3UK-0M72nXEIkfw57MbctVBn_D2X2dk8_K8Wb4V1cfr-_KpKkAbVUiDbtE06ATr-FxYJpErJk2DBpTVzGmJ1kCLesHBWCeM6YC14LiVFgDEjNxfbie4-hj9HuJvfYasJ0jxBxfdTio</recordid><startdate>20200603</startdate><enddate>20200603</enddate><creator>Ni, Minheng</creator><creator>Huang, Haoyang</creator><creator>Su, Lin</creator><creator>Cui, Edward</creator><creator>Bharti, Taroon</creator><creator>Wang, Lijuan</creator><creator>Gao, Jianfeng</creator><creator>Zhang, Dongdong</creator><creator>Duan, Nan</creator><scope>AKY</scope><scope>GOX</scope></search><sort><creationdate>20200603</creationdate><title>M3P: Learning Universal Representations via Multitask Multilingual Multimodal Pre-training</title><author>Ni, Minheng ; Huang, Haoyang ; Su, Lin ; Cui, Edward ; Bharti, Taroon ; Wang, Lijuan ; Gao, Jianfeng ; Zhang, Dongdong ; Duan, Nan</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-LOGICAL-a675-47ec9bbec30f213804e25047be7a5860c64e87ade692a78c377fa0dac2848aaa3</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2020</creationdate><topic>Computer Science - Computation and Language</topic><topic>Computer Science - Computer Vision and Pattern Recognition</topic><toplevel>online_resources</toplevel><creatorcontrib>Ni, Minheng</creatorcontrib><creatorcontrib>Huang, Haoyang</creatorcontrib><creatorcontrib>Su, Lin</creatorcontrib><creatorcontrib>Cui, Edward</creatorcontrib><creatorcontrib>Bharti, Taroon</creatorcontrib><creatorcontrib>Wang, Lijuan</creatorcontrib><creatorcontrib>Gao, Jianfeng</creatorcontrib><creatorcontrib>Zhang, Dongdong</creatorcontrib><creatorcontrib>Duan, Nan</creatorcontrib><collection>arXiv Computer Science</collection><collection>arXiv.org</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext_linktorsrc</fulltext></delivery><addata><au>Ni, Minheng</au><au>Huang, Haoyang</au><au>Su, Lin</au><au>Cui, Edward</au><au>Bharti, Taroon</au><au>Wang, Lijuan</au><au>Gao, Jianfeng</au><au>Zhang, Dongdong</au><au>Duan, Nan</au><format>journal</format><genre>article</genre><ristype>JOUR</ristype><atitle>M3P: Learning Universal Representations via Multitask Multilingual Multimodal Pre-training</atitle><date>2020-06-03</date><risdate>2020</risdate><abstract>We present M3P, a Multitask Multilingual Multimodal Pre-trained model that combines multilingual pre-training and multimodal pre-training into a unified framework via multitask pre-training. Our goal is to learn universal representations that can map objects occurred in different modalities or texts expressed in different languages into a common semantic space. In addition, to explicitly encourage fine-grained alignment between images and non-English languages, we also propose Multimodal Code-switched Training (MCT) to combine monolingual pre-training and multimodal pre-training via a code-switch strategy. Experiments are performed on the multilingual image retrieval task across two benchmark datasets, including MSCOCO and Multi30K. M3P can achieve comparable results for English and new state-of-the-art results for non-English languages.</abstract><doi>10.48550/arxiv.2006.02635</doi><oa>free_for_read</oa></addata></record>
fulltext	fulltext_linktorsrc
identifier	DOI: 10.48550/arxiv.2006.02635
ispartof
issn
language	eng
recordid	cdi_arxiv_primary_2006_02635
source	arXiv.org
subjects	Computer Science - Computation and Language Computer Science - Computer Vision and Pattern Recognition
title	M3P: Learning Universal Representations via Multitask Multilingual Multimodal Pre-training
url	https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-02T18%3A32%3A04IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-arxiv_GOX&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.atitle=M3P:%20Learning%20Universal%20Representations%20via%20Multitask%20Multilingual%20Multimodal%20Pre-training&rft.au=Ni,%20Minheng&rft.date=2020-06-03&rft_id=info:doi/10.48550/arxiv.2006.02635&rft_dat=%3Carxiv_GOX%3E2006_02635%3C/arxiv_GOX%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_id=info:pmid/&rfr_iscdi=true