{"id":"https://openalex.org/W4408355023","doi":"https://doi.org/10.1109/icassp49660.2025.10888427","title":"Language Models Can See Better: Visual Contrastive Decoding For LLM Multimodal Reasoning","display_name":"Language Models Can See Better: Visual Contrastive Decoding For LLM Multimodal Reasoning","publication_year":2025,"publication_date":"2025-03-12","ids":{"openalex":"https://openalex.org/W4408355023","doi":"https://doi.org/10.1109/icassp49660.2025.10888427"},"language":"en","primary_location":{"id":"doi:10.1109/icassp49660.2025.10888427","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49660.2025.10888427","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5035428041","display_name":"Yuqi Pang","orcid":null},"institutions":[{"id":"https://openalex.org/I19820366","display_name":"Chinese Academy of Sciences","ror":"https://ror.org/034t30j35","country_code":"CN","type":"government","lineage":["https://openalex.org/I19820366"]},{"id":"https://openalex.org/I4210156404","display_name":"Institute of Information Engineering","ror":"https://ror.org/04r53se39","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210156404"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuqi Pang","raw_affiliation_strings":["Chinese Academy of Sciences,Institute of Information Engineering,Beijing,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chinese Academy of Sciences,Institute of Information Engineering,Beijing,China","institution_ids":["https://openalex.org/I4210156404","https://openalex.org/I19820366"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5060378291","display_name":"Bowen Yang","orcid":"https://orcid.org/0000-0002-9471-3452"},"institutions":[{"id":"https://openalex.org/I19820366","display_name":"Chinese Academy of Sciences","ror":"https://ror.org/034t30j35","country_code":"CN","type":"government","lineage":["https://openalex.org/I19820366"]},{"id":"https://openalex.org/I4210156404","display_name":"Institute of Information Engineering","ror":"https://ror.org/04r53se39","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210156404"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Bowen Yang","raw_affiliation_strings":["Chinese Academy of Sciences,Institute of Information Engineering,Beijing,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chinese Academy of Sciences,Institute of Information Engineering,Beijing,China","institution_ids":["https://openalex.org/I4210156404","https://openalex.org/I19820366"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5048179343","display_name":"Haoqin Tu","orcid":"https://orcid.org/0000-0002-5627-249X"},"institutions":[{"id":"https://openalex.org/I185103710","display_name":"University of California, Santa Cruz","ror":"https://ror.org/03s65by71","country_code":"US","type":"education","lineage":["https://openalex.org/I185103710"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Haoqin Tu","raw_affiliation_strings":["University of California,Computer and Science Engineering,Santa Cruz,CA,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of California,Computer and Science Engineering,Santa Cruz,CA,USA","institution_ids":["https://openalex.org/I185103710"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5032059481","display_name":"Yun Cao","orcid":"https://orcid.org/0000-0003-3433-0764"},"institutions":[{"id":"https://openalex.org/I19820366","display_name":"Chinese Academy of Sciences","ror":"https://ror.org/034t30j35","country_code":"CN","type":"government","lineage":["https://openalex.org/I19820366"]},{"id":"https://openalex.org/I4210156404","display_name":"Institute of Information Engineering","ror":"https://ror.org/04r53se39","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210156404"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yun Cao","raw_affiliation_strings":["Chinese Academy of Sciences,Institute of Information Engineering,Beijing,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chinese Academy of Sciences,Institute of Information Engineering,Beijing,China","institution_ids":["https://openalex.org/I4210156404","https://openalex.org/I19820366"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100358730","display_name":"Zeyu Zhang","orcid":"https://orcid.org/0000-0002-4035-4028"},"institutions":[{"id":"https://openalex.org/I19820366","display_name":"Chinese Academy of Sciences","ror":"https://ror.org/034t30j35","country_code":"CN","type":"government","lineage":["https://openalex.org/I19820366"]},{"id":"https://openalex.org/I4210156404","display_name":"Institute of Information Engineering","ror":"https://ror.org/04r53se39","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210156404"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zeyu Zhang","raw_affiliation_strings":["Chinese Academy of Sciences,Institute of Information Engineering,Beijing,China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Chinese Academy of Sciences,Institute of Information Engineering,Beijing,China","institution_ids":["https://openalex.org/I4210156404","https://openalex.org/I19820366"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"5"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9988999962806702,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.9988999962806702,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9822999835014343,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10215","display_name":"Semantic Web and Ontologies","score":0.9814000129699707,"subfield":{"id":"https://openalex.org/subfields/1702","display_name":"Artificial Intelligence"},"field":{"id":"https://openalex.org/fields/17","display_name":"Computer Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7418981790542603},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.7340491414070129},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.49269965291023254},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.45673370361328125},{"id":"https://openalex.org/keywords/visual-reasoning","display_name":"Visual reasoning","score":0.4293384850025177},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.3579915463924408}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7418981790542603},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.7340491414070129},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.49269965291023254},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.45673370361328125},{"id":"https://openalex.org/C2777508537","wikidata":"https://www.wikidata.org/wiki/Q7936620","display_name":"Visual reasoning","level":2,"score":0.4293384850025177},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.3579915463924408},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp49660.2025.10888427","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49660.2025.10888427","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":16,"referenced_works":["https://openalex.org/W2560730294","https://openalex.org/W2765716052","https://openalex.org/W2947312908","https://openalex.org/W2963096510","https://openalex.org/W4387426932","https://openalex.org/W4389519587","https://openalex.org/W4389520684","https://openalex.org/W4402727381","https://openalex.org/W4404784276","https://openalex.org/W6791353385","https://openalex.org/W6809646742","https://openalex.org/W6810334672","https://openalex.org/W6811009250","https://openalex.org/W6811013733","https://openalex.org/W6851592950","https://openalex.org/W6856485779"],"related_works":["https://openalex.org/W2161474341","https://openalex.org/W4302615923","https://openalex.org/W3203142394","https://openalex.org/W2351061015","https://openalex.org/W4220731478","https://openalex.org/W1974101135","https://openalex.org/W2017509870","https://openalex.org/W4251141768","https://openalex.org/W4360952157","https://openalex.org/W3204019825"],"abstract_inverted_index":{"Although":[0],"Large":[1,23],"Language":[2,24],"Models":[3,25],"(LLMs)":[4],"excel":[5],"in":[6,148],"reasoning":[7],"and":[8,30,60,83,104,120],"generation":[9],"for":[10,18,69],"language":[11],"tasks,":[12],"they":[13],"are":[14],"not":[15,143],"specifically":[16,67],"designed":[17],"multimodal":[19],"challenges.":[20],"Training":[21],"Multimodal":[22],"(MLLMs),":[26],"however,":[27],"is":[28],"resource-intensive":[29],"constrained":[31],"by":[32,98],"various":[33],"training":[34],"limitations.":[35],"In":[36],"this":[37,50,70],"paper,":[38],"we":[39,91,130],"propose":[40],"the":[41,61,94,123,154],"Modular-based":[42],"Visual":[43],"Contrastive":[44],"Decoding":[45],"(MVCD)":[46],"framework":[47,53],"to":[48,116],"move":[49],"obstacle.":[51],"Our":[52,141,161],"leverages":[54],"LLMs\u2019":[55,113],"In-Context":[56],"Learning":[57],"(ICL)":[58],"capability":[59],"proposed":[62],"visual":[63,79,114],"contrastive-example":[64],"decoding":[65,159],"(CED),":[66],"tailored":[68],"framework,":[71],"without":[72],"requiring":[73],"any":[74],"additional":[75],"training.":[76],"By":[77],"converting":[78],"signals":[80],"into":[81],"text":[82],"focusing":[84],"on":[85,107],"contrastive":[86],"output":[87],"distributions":[88],"during":[89],"decoding,":[90],"can":[92],"highlight":[93],"new":[95],"information":[96],"introduced":[97],"contextual":[99],"examples,":[100],"explore":[101],"their":[102],"connections,":[103],"avoid":[105],"over-reliance":[106],"prior":[108],"encoded":[109],"knowledge.":[110],"MVCD":[111],"enhances":[112],"perception":[115],"make":[117],"it":[118],"see":[119],"reason":[121],"over":[122],"input":[124],"visuals.":[125],"To":[126],"demonstrate":[127],"MVCD\u2019s":[128],"effectiveness,":[129],"conduct":[131],"experiments":[132],"with":[133],"four":[134],"LLMs":[135],"across":[136],"five":[137],"question":[138],"answering":[139],"datasets.":[140],"results":[142],"only":[144],"show":[145],"consistent":[146],"improvement":[147],"model":[149],"accuracy":[150],"but":[151],"well":[152],"explain":[153],"effective":[155],"components":[156],"inside":[157],"our":[158],"strategy.":[160],"code":[162],"will":[163],"be":[164],"available":[165],"at":[166],"https://github.com/Pbhgit/MVCD.":[167]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":1}],"updated_date":"2026-07-16T13:24:37.021932","created_date":"2025-10-10T00:00:00"}