{"id":"https://openalex.org/W4414360528","doi":"https://doi.org/10.24963/ijcai.2025/46","title":"Modality-Fair Preference Optimization for Trustworthy MLLM Alignment","display_name":"Modality-Fair Preference Optimization for Trustworthy MLLM Alignment","publication_year":2025,"publication_date":"2025-09-01","ids":{"openalex":"https://openalex.org/W4414360528","doi":"https://doi.org/10.24963/ijcai.2025/46"},"language":"en","primary_location":{"id":"doi:10.24963/ijcai.2025/46","is_oa":false,"landing_page_url":"https://doi.org/10.24963/ijcai.2025/46","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence","raw_type":"proceedings-article"},"type":"article","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/A5113656054","display_name":"Songtao Jiang","orcid":null},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Songtao Jiang","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100456377","display_name":"Yan Zhang","orcid":"https://orcid.org/0000-0003-1585-0801"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yan Zhang","raw_affiliation_strings":["ByteDance"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ByteDance","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5117428676","display_name":"Ruizhe Chen","orcid":"https://orcid.org/0000-0003-2302-6775"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Ruizhe Chen","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5033563561","display_name":"Tianxiang Hu","orcid":"https://orcid.org/0000-0001-9897-8098"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Tianxiang Hu","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5040632841","display_name":"Yeying Jin","orcid":"https://orcid.org/0000-0001-7818-9534"},"institutions":[{"id":"https://openalex.org/I165932596","display_name":"National University of Singapore","ror":"https://ror.org/01tgyzw49","country_code":"SG","type":"education","lineage":["https://openalex.org/I165932596"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Yeying Jin","raw_affiliation_strings":["National University of Singapore"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"National University of Singapore","institution_ids":["https://openalex.org/I165932596"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101357077","display_name":"Qinglin He","orcid":null},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Qinglin He","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101313656","display_name":"Feng Yang","orcid":"https://orcid.org/0009-0007-2409-926X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yang Feng","raw_affiliation_strings":["Angelalign Inc., China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Angelalign Inc., China","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101913769","display_name":"Jian Wu","orcid":"https://orcid.org/0000-0003-4518-7027"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jian Wu","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5024343415","display_name":"Zuozhu Liu","orcid":"https://orcid.org/0000-0002-7816-502X"},"institutions":[{"id":"https://openalex.org/I76130692","display_name":"Zhejiang University","ror":"https://ror.org/00a2xv884","country_code":"CN","type":"education","lineage":["https://openalex.org/I76130692"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zuozhu Liu","raw_affiliation_strings":["Zhejiang University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Zhejiang University","institution_ids":["https://openalex.org/I76130692"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":9,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.31429025,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"403","last_page":"411"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10927","display_name":"Access Control and Trust","score":0.9832000136375427,"subfield":{"id":"https://openalex.org/subfields/3312","display_name":"Sociology and Political Science"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},"topics":[{"id":"https://openalex.org/T10927","display_name":"Access Control and Trust","score":0.9832000136375427,"subfield":{"id":"https://openalex.org/subfields/3312","display_name":"Sociology and Political Science"},"field":{"id":"https://openalex.org/fields/33","display_name":"Social Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T10456","display_name":"Multi-Agent Systems and Negotiation","score":0.9643999934196472,"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/leverage","display_name":"Leverage (statistics)","score":0.6642000079154968},{"id":"https://openalex.org/keywords/trustworthiness","display_name":"Trustworthiness","score":0.6402000188827515},{"id":"https://openalex.org/keywords/context","display_name":"Context (archaeology)","score":0.5493999719619751},{"id":"https://openalex.org/keywords/preference","display_name":"Preference","score":0.5329999923706055},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.5085999965667725},{"id":"https://openalex.org/keywords/structured-prediction","display_name":"Structured prediction","score":0.438400000333786},{"id":"https://openalex.org/keywords/preference-learning","display_name":"Preference learning","score":0.4180999994277954},{"id":"https://openalex.org/keywords/object","display_name":"Object (grammar)","score":0.4104999899864197}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7099999785423279},{"id":"https://openalex.org/C153083717","wikidata":"https://www.wikidata.org/wiki/Q6535263","display_name":"Leverage (statistics)","level":2,"score":0.6642000079154968},{"id":"https://openalex.org/C153701036","wikidata":"https://www.wikidata.org/wiki/Q659974","display_name":"Trustworthiness","level":2,"score":0.6402000188827515},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.5493999719619751},{"id":"https://openalex.org/C2781249084","wikidata":"https://www.wikidata.org/wiki/Q908656","display_name":"Preference","level":2,"score":0.5329999923706055},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5180000066757202},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5085999965667725},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.438400000333786},{"id":"https://openalex.org/C181204326","wikidata":"https://www.wikidata.org/wiki/Q7239820","display_name":"Preference learning","level":3,"score":0.4180999994277954},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4124999940395355},{"id":"https://openalex.org/C2781238097","wikidata":"https://www.wikidata.org/wiki/Q175026","display_name":"Object (grammar)","level":2,"score":0.4104999899864197},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.40779998898506165},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.39079999923706055},{"id":"https://openalex.org/C2780522230","wikidata":"https://www.wikidata.org/wiki/Q1140419","display_name":"Ambiguity","level":2,"score":0.3458000123500824},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.30559998750686646},{"id":"https://openalex.org/C64754055","wikidata":"https://www.wikidata.org/wiki/Q7574053","display_name":"Spatial contextual awareness","level":2,"score":0.29809999465942383},{"id":"https://openalex.org/C64876066","wikidata":"https://www.wikidata.org/wiki/Q5141226","display_name":"Cognitive neuroscience of visual object recognition","level":3,"score":0.2904999852180481},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.28029999136924744},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.2754000127315521},{"id":"https://openalex.org/C2780226545","wikidata":"https://www.wikidata.org/wiki/Q6888030","display_name":"Modality (human\u2013computer interaction)","level":2,"score":0.27239999175071716},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.27000001072883606},{"id":"https://openalex.org/C2776502983","wikidata":"https://www.wikidata.org/wiki/Q690182","display_name":"Contrast (vision)","level":2,"score":0.2653000056743622},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.25780001282691956},{"id":"https://openalex.org/C26760741","wikidata":"https://www.wikidata.org/wiki/Q160402","display_name":"Perception","level":2,"score":0.2558000087738037}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.24963/ijcai.2025/46","is_oa":false,"landing_page_url":"https://doi.org/10.24963/ijcai.2025/46","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Multimodal":[0],"large":[1],"language":[2],"models":[3,32,83,217],"(MLLMs)":[4],"have":[5],"achieved":[6],"remarkable":[7],"success":[8],"across":[9],"various":[10],"tasks.":[11],"However,":[12],"separate":[13],"training":[14],"of":[15,25,55,77,98,152,209,229],"visual":[16,99],"and":[17,185,233],"textual":[18],"encoders":[19],"often":[20],"results":[21],"in":[22,57,119,157,165],"a":[23,43,125,153],"misalignment":[24,29,126],"the":[26,40,53,75,91,105,110,115,120,128,150,174,182,207,215,230],"modality.":[27],"Such":[28],"may":[30],"lead":[31],"to":[33,46,62,66,85,176,191,218],"generate":[34,177],"content":[35],"that":[36,74,104,203],"is":[37,94],"absent":[38],"from":[39,162],"input":[41,92,134,183],"image,":[42],"phenomenon":[44],"referred":[45],"as":[47],"hallucination.":[48],"These":[49,122],"inaccuracies":[50],"severely":[51],"undermine":[52],"trustworthiness":[54,76,200,208,220],"MLLMs":[56,78],"real-world":[58],"applications.":[59],"Despite":[60],"attempts":[61],"optimize":[63],"text":[64],"preferences":[65],"mitigate":[67],"this":[68],"issue,":[69],"our":[70,138],"initial":[71],"investigation":[72],"indicates":[73,103],"remains":[79],"inadequate.":[80],"Specifically,":[81],"these":[82],"tend":[84],"provide":[86],"preferred":[87],"answers":[88,131,178],"even":[89,226],"when":[90],"image":[93,169],"heavily":[95],"distorted.":[96],"Analysis":[97],"token":[100],"attention":[101],"also":[102],"model":[106,175],"focuses":[107],"primarily":[108],"on":[109,198,222],"surrounding":[111],"context":[112],"rather":[113],"than":[114],"key":[116,166],"object":[117],"referenced":[118],"question.":[121],"findings":[123],"highlight":[124],"between":[127],"modalities,":[129],"where":[130],"inadequately":[132],"leverage":[133],"images.":[135],"Motivated":[136],"by":[137],"findings,":[139],"we":[140],"propose":[141],"Modality-Fair":[142],"Preference":[143],"Optimization":[144],"(MFPO),":[145],"which":[146,158],"comprises":[147],"three":[148,199],"components:":[149],"construction":[151],"multimodal":[154],"preference":[155],"dataset":[156],"dispreferred":[159],"images":[160],"differ":[161],"originals":[163],"solely":[164],"regions;":[167],"an":[168,186],"reward":[170],"loss":[171],"function":[172],"encouraging":[173],"better":[179],"aligned":[180],"with":[181],"images;":[184],"easy-to-hard":[187],"iterative":[188],"alignment":[189],"strategy":[190],"stabilize":[192],"joint":[193],"modality":[194],"training.":[195],"Extensive":[196],"experiments":[197],"benchmarks":[201],"demonstrate":[202],"MFPO":[204],"significantly":[205],"enhances":[206],"MLLMs.":[210],"In":[211],"particular,":[212],"it":[213],"enables":[214],"7B":[216],"attain":[219],"levels":[221],"par":[223],"with,":[224],"or":[225],"surpass,":[227],"those":[228],"13B,":[231],"34B,":[232],"larger":[234],"models.":[235]},"counts_by_year":[],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}