Retrieve, Caption, Generate: Visual Grounding for Enhancing Commonsense in Text Generation Models

Retrieve, Caption, Generate: Visual Grounding for Enhancing Commonsense in Text Generation Models

We investigate the use of multimodal information contained in images as an effective method for enhancing the commonsense of Transformer models for text generation. We perform experiments using BART and T5 on concept-to-text generation, specifically the task of generative commonsense reasoning, or C...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:arXiv.org 2022-03
Hauptverfasser: Feng, Steven Y, Lu, Kevin, Zhuofu Tao, Alikhani, Malihe, Mitamura, Teruko, Hovy, Eduard, Gangal, Varun
Format: Artikel
Sprache:eng
Schlagworte:
Image enhancement
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 Feng, Steven Y
Lu, Kevin
Zhuofu Tao
Alikhani, Malihe
Mitamura, Teruko
Hovy, Eduard
Gangal, Varun
description We investigate the use of multimodal information contained in images as an effective method for enhancing the commonsense of Transformer models for text generation. We perform experiments using BART and T5 on concept-to-text generation, specifically the task of generative commonsense reasoning, or CommonGen. We call our approach VisCTG: Visually Grounded Concept-to-Text Generation. VisCTG involves captioning images representing appropriate everyday scenarios, and using these captions to enrich and steer the generation process. Comprehensive evaluation and analysis demonstrate that VisCTG noticeably improves model performance while successfully addressing several issues of the baseline generations, including poor commonsense, fluency, and specificity.
format Article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2571338957</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2571338957</sourcerecordid><originalsourceid>FETCH-proquest_journals_25713389573</originalsourceid><addsrcrecordid>eNqNjE8LgkAUxJcgSMrv8KCrgu5mVlcxu3QJ6SpLvmrD3tr-iT5-BnUPBobhNzMjFnAh0ni14HzCQmtvSZLwZc6zTARMHtAZhU-MoJC9U5oiqJDQSIcbOCrrZQeV0Z5aRRc4awMlXSWdPqnQ97smi4NAEdT4cr_xcAR73WJnZ2x8lp3F8OtTNt-WdbGLe6MfHq1rbtobGlDDszwVYrXOcvFf6w3sEkTh</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2571338957</pqid></control><display><type>article</type><title>Retrieve, Caption, Generate: Visual Grounding for Enhancing Commonsense in Text Generation Models</title><source>Free E- Journals</source><creator>Feng, Steven Y ; Lu, Kevin ; Zhuofu Tao ; Alikhani, Malihe ; Mitamura, Teruko ; Hovy, Eduard ; Gangal, Varun</creator><creatorcontrib>Feng, Steven Y ; Lu, Kevin ; Zhuofu Tao ; Alikhani, Malihe ; Mitamura, Teruko ; Hovy, Eduard ; Gangal, Varun</creatorcontrib><description>We investigate the use of multimodal information contained in images as an effective method for enhancing the commonsense of Transformer models for text generation. We perform experiments using BART and T5 on concept-to-text generation, specifically the task of generative commonsense reasoning, or CommonGen. We call our approach VisCTG: Visually Grounded Concept-to-Text Generation. VisCTG involves captioning images representing appropriate everyday scenarios, and using these captions to enrich and steer the generation process. Comprehensive evaluation and analysis demonstrate that VisCTG noticeably improves model performance while successfully addressing several issues of the baseline generations, including poor commonsense, fluency, and specificity.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Image enhancement</subject><ispartof>arXiv.org, 2022-03</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>Feng, Steven Y</creatorcontrib><creatorcontrib>Lu, Kevin</creatorcontrib><creatorcontrib>Zhuofu Tao</creatorcontrib><creatorcontrib>Alikhani, Malihe</creatorcontrib><creatorcontrib>Mitamura, Teruko</creatorcontrib><creatorcontrib>Hovy, Eduard</creatorcontrib><creatorcontrib>Gangal, Varun</creatorcontrib><title>Retrieve, Caption, Generate: Visual Grounding for Enhancing Commonsense in Text Generation Models</title><title>arXiv.org</title><description>We investigate the use of multimodal information contained in images as an effective method for enhancing the commonsense of Transformer models for text generation. We perform experiments using BART and T5 on concept-to-text generation, specifically the task of generative commonsense reasoning, or CommonGen. We call our approach VisCTG: Visually Grounded Concept-to-Text Generation. VisCTG involves captioning images representing appropriate everyday scenarios, and using these captions to enrich and steer the generation process. Comprehensive evaluation and analysis demonstrate that VisCTG noticeably improves model performance while successfully addressing several issues of the baseline generations, including poor commonsense, fluency, and specificity.</description><subject>Image enhancement</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2022</creationdate><recordtype>article</recordtype><sourceid>BENPR</sourceid><recordid>eNqNjE8LgkAUxJcgSMrv8KCrgu5mVlcxu3QJ6SpLvmrD3tr-iT5-BnUPBobhNzMjFnAh0ni14HzCQmtvSZLwZc6zTARMHtAZhU-MoJC9U5oiqJDQSIcbOCrrZQeV0Z5aRRc4awMlXSWdPqnQ97smi4NAEdT4cr_xcAR73WJnZ2x8lp3F8OtTNt-WdbGLe6MfHq1rbtobGlDDszwVYrXOcvFf6w3sEkTh</recordid><startdate>20220325</startdate><enddate>20220325</enddate><creator>Feng, Steven Y</creator><creator>Lu, Kevin</creator><creator>Zhuofu Tao</creator><creator>Alikhani, Malihe</creator><creator>Mitamura, Teruko</creator><creator>Hovy, Eduard</creator><creator>Gangal, Varun</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>20220325</creationdate><title>Retrieve, Caption, Generate: Visual Grounding for Enhancing Commonsense in Text Generation Models</title><author>Feng, Steven Y ; Lu, Kevin ; Zhuofu Tao ; Alikhani, Malihe ; Mitamura, Teruko ; Hovy, Eduard ; Gangal, Varun</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_25713389573</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2022</creationdate><topic>Image enhancement</topic><toplevel>online_resources</toplevel><creatorcontrib>Feng, Steven Y</creatorcontrib><creatorcontrib>Lu, Kevin</creatorcontrib><creatorcontrib>Zhuofu Tao</creatorcontrib><creatorcontrib>Alikhani, Malihe</creatorcontrib><creatorcontrib>Mitamura, Teruko</creatorcontrib><creatorcontrib>Hovy, Eduard</creatorcontrib><creatorcontrib>Gangal, Varun</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; 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>Feng, Steven Y</au><au>Lu, Kevin</au><au>Zhuofu Tao</au><au>Alikhani, Malihe</au><au>Mitamura, Teruko</au><au>Hovy, Eduard</au><au>Gangal, Varun</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Retrieve, Caption, Generate: Visual Grounding for Enhancing Commonsense in Text Generation Models</atitle><jtitle>arXiv.org</jtitle><date>2022-03-25</date><risdate>2022</risdate><eissn>2331-8422</eissn><abstract>We investigate the use of multimodal information contained in images as an effective method for enhancing the commonsense of Transformer models for text generation. We perform experiments using BART and T5 on concept-to-text generation, specifically the task of generative commonsense reasoning, or CommonGen. We call our approach VisCTG: Visually Grounded Concept-to-Text Generation. VisCTG involves captioning images representing appropriate everyday scenarios, and using these captions to enrich and steer the generation process. Comprehensive evaluation and analysis demonstrate that VisCTG noticeably improves model performance while successfully addressing several issues of the baseline generations, including poor commonsense, fluency, and specificity.</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, 2022-03
issn 2331-8422
language eng
recordid cdi_proquest_journals_2571338957
source Free E- Journals
subjects Image enhancement
title Retrieve, Caption, Generate: Visual Grounding for Enhancing Commonsense in Text Generation Models
url https://sfx.bib-bvb.de/sfx_tum?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-02-02T04%3A27%3A10IST&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=Retrieve,%20Caption,%20Generate:%20Visual%20Grounding%20for%20Enhancing%20Commonsense%20in%20Text%20Generation%20Models&rft.jtitle=arXiv.org&rft.au=Feng,%20Steven%20Y&rft.date=2022-03-25&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2571338957%3C/proquest%3E%3Curl%3E%3C/url%3E&disable_directlink=true&sfx.directlink=off&sfx.report_link=0&rft_id=info:oai/&rft_pqid=2571338957&rft_id=info:pmid/&rfr_iscdi=true