{"id":"https://openalex.org/W7138318110","doi":"https://doi.org/10.1609/aaai.v40i27.39405","title":"EvoMoE: Expert Evolution in Mixture of Experts for Multimodal Large Language Models","display_name":"EvoMoE: Expert Evolution in Mixture of Experts for Multimodal Large Language Models","publication_year":2026,"publication_date":"2026-03-14","ids":{"openalex":"https://openalex.org/W7138318110","doi":"https://doi.org/10.1609/aaai.v40i27.39405"},"language":null,"primary_location":{"id":"doi:10.1609/aaai.v40i27.39405","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i27.39405","pdf_url":null,"source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://doi.org/10.1609/aaai.v40i27.39405","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5044802124","display_name":"Linglin Jing","orcid":"https://orcid.org/0000-0003-1287-4066"},"institutions":[{"id":"https://openalex.org/I4391012619","display_name":"Shanghai Artificial Intelligence Laboratory","ror":"https://ror.org/03wkvpx79","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4391012619"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Linglin Jing","raw_affiliation_strings":["Shanghai AI Lab"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai AI Lab","institution_ids":["https://openalex.org/I4391012619"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129675768","display_name":"Yuting Gao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Yuting Gao","raw_affiliation_strings":["Ant Group"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Ant Group","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129682684","display_name":"Zhigang Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I4391012619","display_name":"Shanghai Artificial Intelligence Laboratory","ror":"https://ror.org/03wkvpx79","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4391012619"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhigang Wang","raw_affiliation_strings":["Shanghai AI Lab"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai AI Lab","institution_ids":["https://openalex.org/I4391012619"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129721331","display_name":"Wang Lan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang Lan","raw_affiliation_strings":["Ant Group"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Ant Group","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129672041","display_name":"Yiwen Tang","orcid":null},"institutions":[{"id":"https://openalex.org/I4391012619","display_name":"Shanghai Artificial Intelligence Laboratory","ror":"https://ror.org/03wkvpx79","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4391012619"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yiwen Tang","raw_affiliation_strings":["Shanghai AI Lab"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai AI Lab","institution_ids":["https://openalex.org/I4391012619"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129729802","display_name":"Weiyun Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I4391012619","display_name":"Shanghai Artificial Intelligence Laboratory","ror":"https://ror.org/03wkvpx79","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4391012619"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Weiyun Wang","raw_affiliation_strings":["Shanghai AI Lab"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai AI Lab","institution_ids":["https://openalex.org/I4391012619"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129730176","display_name":"Wenhai Wang","orcid":null},"institutions":[{"id":"https://openalex.org/I4391012619","display_name":"Shanghai Artificial Intelligence Laboratory","ror":"https://ror.org/03wkvpx79","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4391012619"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wenhai Wang","raw_affiliation_strings":["Shanghai AI Lab"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Shanghai AI Lab","institution_ids":["https://openalex.org/I4391012619"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5129704625","display_name":"Qingpei Guo","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qingpei Guo","raw_affiliation_strings":["Ant Group"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Ant Group","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":"40","issue":"27","first_page":"22463","last_page":"22471"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11714","display_name":"Multimodal Machine Learning Applications","score":0.767799973487854,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.767799973487854,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T10036","display_name":"Advanced Neural Network Applications","score":0.031300000846385956,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.027300000190734863,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/router","display_name":"Router","score":0.6929000020027161},{"id":"https://openalex.org/keywords/expert-system","display_name":"Expert system","score":0.5755000114440918},{"id":"https://openalex.org/keywords/initialization","display_name":"Initialization","score":0.48030000925064087},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.4528000056743622},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.423799991607666},{"id":"https://openalex.org/keywords/subject-matter-expert","display_name":"Subject-matter expert","score":0.39989998936653137},{"id":"https://openalex.org/keywords/routing","display_name":"Routing (electronic design automation)","score":0.3813999891281128}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7748000025749207},{"id":"https://openalex.org/C2775896111","wikidata":"https://www.wikidata.org/wiki/Q642560","display_name":"Router","level":2,"score":0.6929000020027161},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5773000121116638},{"id":"https://openalex.org/C58328972","wikidata":"https://www.wikidata.org/wiki/Q184609","display_name":"Expert system","level":2,"score":0.5755000114440918},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5188000202178955},{"id":"https://openalex.org/C114466953","wikidata":"https://www.wikidata.org/wiki/Q6034165","display_name":"Initialization","level":2,"score":0.48030000925064087},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.4528000056743622},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.423799991607666},{"id":"https://openalex.org/C105002631","wikidata":"https://www.wikidata.org/wiki/Q4833645","display_name":"Subject-matter expert","level":3,"score":0.39989998936653137},{"id":"https://openalex.org/C74172769","wikidata":"https://www.wikidata.org/wiki/Q1446839","display_name":"Routing (electronic design automation)","level":2,"score":0.3813999891281128},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.35420000553131104},{"id":"https://openalex.org/C136197465","wikidata":"https://www.wikidata.org/wiki/Q1729295","display_name":"Variety (cybernetics)","level":2,"score":0.29580000042915344},{"id":"https://openalex.org/C132525143","wikidata":"https://www.wikidata.org/wiki/Q141488","display_name":"Graph","level":2,"score":0.28999999165534973},{"id":"https://openalex.org/C112313634","wikidata":"https://www.wikidata.org/wiki/Q7886648","display_name":"Complement (music)","level":5,"score":0.289000004529953},{"id":"https://openalex.org/C2780598303","wikidata":"https://www.wikidata.org/wiki/Q65921492","display_name":"Flexibility (engineering)","level":2,"score":0.2809999883174896},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.26440000534057617},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.26339998841285706},{"id":"https://openalex.org/C102600418","wikidata":"https://www.wikidata.org/wiki/Q6517507","display_name":"Legal expert system","level":3,"score":0.2500999867916107}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1609/aaai.v40i27.39405","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i27.39405","pdf_url":null,"source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1609/aaai.v40i27.39405","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i27.39405","pdf_url":null,"source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"sustainable_development_goals":[{"display_name":"Quality Education","id":"https://metadata.un.org/sdg/4","score":0.44146600365638733}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Recent":[0],"advancements":[1],"have":[2,36],"shown":[3],"that":[4,141,157,175,210],"the":[5,13,69,81,85,93,166,232,238,244],"Mixture":[6],"of":[7,15,84,96,220,234,240,247],"Experts":[8],"(MoE)":[9],"approach":[10],"significantly":[11,212],"enhances":[12],"capacity":[14],"large":[16,32],"language":[17,33],"models":[18,34],"(LLMs)":[19],"and":[20,54,79,109,119,186,227,250],"improves":[21],"performance":[22,239],"on":[23,27,183],"downstream":[24],"tasks.":[25],"Building":[26],"these":[28,123],"promising":[29],"results,":[30],"multi-modal":[31,43,221],"(MLLMs)":[35],"increasingly":[37],"adopted":[38],"MoE":[39,44,61,86,130],"techniques.":[40],"However,":[41],"existing":[42],"tuning":[45,131],"methods":[46],"typically":[47],"face":[48],"two":[49],"key":[50],"challenges:":[51],"expert":[52,77,101,115,138,155,161,248],"uniformity":[53,58,249],"router":[55,89,251],"rigidity.":[56,252],"Expert":[57],"occurs":[59],"because":[60],"experts":[62,146,181],"are":[63],"often":[64],"initialized":[65],"by":[66,195,242],"simply":[67],"replicating":[68],"FFN":[70],"parameters":[71],"from":[72,92,147],"LLMs,":[73],"leading":[74],"to":[75,105,179],"homogenized":[76],"functions":[78],"weakening":[80],"intended":[82],"diversification":[83],"architecture.":[87],"Meanwhile,":[88],"rigidity":[90],"stems":[91],"prevalent":[94],"use":[95],"static":[97],"linear":[98],"routers":[99],"for":[100,117,203],"selection,":[102],"which":[103,197],"fail":[104],"distinguish":[106],"between":[107],"visual":[108],"textual":[110],"tokens,":[111],"resulting":[112],"in":[113,236],"similar":[114],"distributions":[116],"image":[118],"text.":[120],"To":[121],"address":[122],"limitations,":[124],"we":[125,164],"propose":[126],"EvoMoE,":[127],"an":[128],"innovative":[129],"framework.":[132],"EvoMoE":[133,211,235],"introduces":[134],"a":[135,148,152,171,218],"meticulously":[136],"designed":[137],"initialization":[139],"strategy":[140],"progressively":[142],"evolves":[143],"multiple":[144],"robust":[145],"single":[149],"trainable":[150],"expert,":[151],"process":[153],"termed":[154],"evolution":[156],"specifically":[158],"targets":[159],"severe":[160],"homogenization.":[162],"Furthermore,":[163],"introduce":[165],"Dynamic":[167],"Token-aware":[168],"Router":[169],"(DTR),":[170],"novel":[172],"routing":[173,192,200],"mechanism":[174],"allocates":[176],"input":[177],"tokens":[178],"appropriate":[180],"based":[182],"their":[184],"modality":[185],"intrinsic":[187],"token":[188],"values.":[189],"This":[190],"dynamic":[191],"is":[193],"facilitated":[194],"hypernetworks,":[196],"dynamically":[198],"generate":[199],"weights":[201],"tailored":[202],"each":[204],"individual":[205],"token.":[206],"Extensive":[207],"experiments":[208],"demonstrate":[209],"outperforms":[213],"other":[214],"sparse":[215],"MLLMs":[216,241],"across":[217],"variety":[219],"benchmarks,":[222],"including":[223],"MME,":[224],"MMBench,":[225],"TextVQA,":[226],"POPE.":[228],"Our":[229],"results":[230],"highlight":[231],"effectiveness":[233],"enhancing":[237],"addressing":[243],"critical":[245],"issues":[246]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-07-22T07:51:19.307946","created_date":"2026-03-18T00:00:00"}