Large-scale Bilingual Language-Image Contrastive Learning
This paper is a technical report to share our experience and findings building a Korean and English bilingual multimodal model. While many of the multimodal datasets focus on English and multilingual multimodal research uses machine-translated texts, employing such machine-translated texts is limite...
Gespeichert in:
Veröffentlicht in: | arXiv.org 2022-04 |
---|---|
Hauptverfasser: | , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
container_end_page | |
---|---|
container_issue | |
container_start_page | |
container_title | arXiv.org |
container_volume | |
creator | Ko, Byungsoo Gu, Geonmo |
description | This paper is a technical report to share our experience and findings building a Korean and English bilingual multimodal model. While many of the multimodal datasets focus on English and multilingual multimodal research uses machine-translated texts, employing such machine-translated texts is limited to describing unique expressions, cultural information, and proper noun in languages other than English. In this work, we collect 1.1 billion image-text pairs (708 million Korean and 476 million English) and train a bilingual multimodal model named KELIP. We introduce simple yet effective training schemes, including MAE pre-training and multi-crop augmentation. Extensive experiments demonstrate that a model trained with such training schemes shows competitive performance in both languages. Moreover, we discuss multimodal-related research questions: 1) strong augmentation-based methods can distract the model from learning proper multimodal relations; 2) training multimodal model without cross-lingual relation can learn the relation via visual semantics; 3) our bilingual KELIP can capture cultural differences of visual semantics for the same meaning of words; 4) a large-scale multimodal model can be used for multimodal feature analogy. We hope that this work will provide helpful experience and findings for future research. We provide an open-source pre-trained KELIP. |
format | Article |
fullrecord | <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2644595309</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2644595309</sourcerecordid><originalsourceid>FETCH-proquest_journals_26445953093</originalsourceid><addsrcrecordid>eNqNiksKwjAUAIMgWLR3CLgOxHyq2VoUhSzdl0eJJSUmmo_nN4IHcDOzmFmghnG-IwfB2Aq1Kc2UUtbtmZS8QUpDnAxJIziDj9ZZPxVwWMPXNVwflbgPPkdI2b4N1gair9sGLe_gkml_XqPt-XTrL-QZw6uYlIc5lOhrGlgnhFSSU8X_uz5jJzZJ</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2644595309</pqid></control><display><type>article</type><title>Large-scale Bilingual Language-Image Contrastive Learning</title><source>Free E- Journals</source><creator>Ko, Byungsoo ; Gu, Geonmo</creator><creatorcontrib>Ko, Byungsoo ; Gu, Geonmo</creatorcontrib><description>This paper is a technical report to share our experience and findings building a Korean and English bilingual multimodal model. While many of the multimodal datasets focus on English and multilingual multimodal research uses machine-translated texts, employing such machine-translated texts is limited to describing unique expressions, cultural information, and proper noun in languages other than English. In this work, we collect 1.1 billion image-text pairs (708 million Korean and 476 million English) and train a bilingual multimodal model named KELIP. We introduce simple yet effective training schemes, including MAE pre-training and multi-crop augmentation. Extensive experiments demonstrate that a model trained with such training schemes shows competitive performance in both languages. Moreover, we discuss multimodal-related research questions: 1) strong augmentation-based methods can distract the model from learning proper multimodal relations; 2) training multimodal model without cross-lingual relation can learn the relation via visual semantics; 3) our bilingual KELIP can capture cultural differences of visual semantics for the same meaning of words; 4) a large-scale multimodal model can be used for multimodal feature analogy. We hope that this work will provide helpful experience and findings for future research. We provide an open-source pre-trained KELIP.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Augmentation ; Bilingualism ; Cultural factors ; English language ; Languages ; Learning ; Machine translation ; Non-English languages ; Semantics ; Texts ; Training</subject><ispartof>arXiv.org, 2022-04</ispartof><rights>2022. This work is published under http://arxiv.org/licenses/nonexclusive-distrib/1.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</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>776,780</link.rule.ids></links><search><creatorcontrib>Ko, Byungsoo</creatorcontrib><creatorcontrib>Gu, Geonmo</creatorcontrib><title>Large-scale Bilingual Language-Image Contrastive Learning</title><title>arXiv.org</title><description>This paper is a technical report to share our experience and findings building a Korean and English bilingual multimodal model. While many of the multimodal datasets focus on English and multilingual multimodal research uses machine-translated texts, employing such machine-translated texts is limited to describing unique expressions, cultural information, and proper noun in languages other than English. In this work, we collect 1.1 billion image-text pairs (708 million Korean and 476 million English) and train a bilingual multimodal model named KELIP. We introduce simple yet effective training schemes, including MAE pre-training and multi-crop augmentation. Extensive experiments demonstrate that a model trained with such training schemes shows competitive performance in both languages. Moreover, we discuss multimodal-related research questions: 1) strong augmentation-based methods can distract the model from learning proper multimodal relations; 2) training multimodal model without cross-lingual relation can learn the relation via visual semantics; 3) our bilingual KELIP can capture cultural differences of visual semantics for the same meaning of words; 4) a large-scale multimodal model can be used for multimodal feature analogy. We hope that this work will provide helpful experience and findings for future research. We provide an open-source pre-trained KELIP.</description><subject>Augmentation</subject><subject>Bilingualism</subject><subject>Cultural factors</subject><subject>English language</subject><subject>Languages</subject><subject>Learning</subject><subject>Machine translation</subject><subject>Non-English languages</subject><subject>Semantics</subject><subject>Texts</subject><subject>Training</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>BENPR</sourceid><recordid>eNqNiksKwjAUAIMgWLR3CLgOxHyq2VoUhSzdl0eJJSUmmo_nN4IHcDOzmFmghnG-IwfB2Aq1Kc2UUtbtmZS8QUpDnAxJIziDj9ZZPxVwWMPXNVwflbgPPkdI2b4N1gair9sGLe_gkml_XqPt-XTrL-QZw6uYlIc5lOhrGlgnhFSSU8X_uz5jJzZJ</recordid><startdate>20220415</startdate><enddate>20220415</enddate><creator>Ko, Byungsoo</creator><creator>Gu, Geonmo</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PRINS</scope><scope>PTHSS</scope></search><sort><creationdate>20220415</creationdate><title>Large-scale Bilingual Language-Image Contrastive Learning</title><author>Ko, Byungsoo ; Gu, Geonmo</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_26445953093</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Augmentation</topic><topic>Bilingualism</topic><topic>Cultural factors</topic><topic>English language</topic><topic>Languages</topic><topic>Learning</topic><topic>Machine translation</topic><topic>Non-English languages</topic><topic>Semantics</topic><topic>Texts</topic><topic>Training</topic><toplevel>online_resources</toplevel><creatorcontrib>Ko, Byungsoo</creatorcontrib><creatorcontrib>Gu, Geonmo</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science & Engineering Collection</collection><collection>ProQuest Central (Alumni Edition)</collection><collection>ProQuest Central UK/Ireland</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central Korea</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>ProQuest Central China</collection><collection>Engineering Collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Ko, Byungsoo</au><au>Gu, Geonmo</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Large-scale Bilingual Language-Image Contrastive Learning</atitle><jtitle>arXiv.org</jtitle><date>2022-04-15</date><risdate>2022</risdate><eissn>2331-8422</eissn><abstract>This paper is a technical report to share our experience and findings building a Korean and English bilingual multimodal model. While many of the multimodal datasets focus on English and multilingual multimodal research uses machine-translated texts, employing such machine-translated texts is limited to describing unique expressions, cultural information, and proper noun in languages other than English. In this work, we collect 1.1 billion image-text pairs (708 million Korean and 476 million English) and train a bilingual multimodal model named KELIP. We introduce simple yet effective training schemes, including MAE pre-training and multi-crop augmentation. Extensive experiments demonstrate that a model trained with such training schemes shows competitive performance in both languages. Moreover, we discuss multimodal-related research questions: 1) strong augmentation-based methods can distract the model from learning proper multimodal relations; 2) training multimodal model without cross-lingual relation can learn the relation via visual semantics; 3) our bilingual KELIP can capture cultural differences of visual semantics for the same meaning of words; 4) a large-scale multimodal model can be used for multimodal feature analogy. We hope that this work will provide helpful experience and findings for future research. We provide an open-source pre-trained KELIP.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record> |
fulltext | fulltext |
identifier | EISSN: 2331-8422 |
ispartof | arXiv.org, 2022-04 |
issn | 2331-8422 |
language | eng |
recordid | cdi_proquest_journals_2644595309 |
source | Free E- Journals |
subjects | Augmentation Bilingualism Cultural factors English language Languages Learning Machine translation Non-English languages Semantics Texts Training |
title | Large-scale Bilingual Language-Image Contrastive Learning |
url | https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-04T00%3A41%3A33IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=Large-scale%20Bilingual%20Language-Image%20Contrastive%20Learning&rft.jtitle=arXiv.org&rft.au=Ko,%20Byungsoo&rft.date=2022-04-15&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2644595309%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2644595309&rft_id=info:pmid/&rfr_iscdi=true |