{"id":"https://openalex.org/W7140215383","doi":"https://doi.org/10.48550/arxiv.2603.21276","title":"Aggregation Alignment for Federated Learning with Mixture-of-Experts under Data Heterogeneity","display_name":"Aggregation Alignment for Federated Learning with Mixture-of-Experts under Data Heterogeneity","publication_year":2026,"publication_date":"2026-03-22","ids":{"openalex":"https://openalex.org/W7140215383","doi":"https://doi.org/10.48550/arxiv.2603.21276"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.21276","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.21276","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2603.21276","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Fang, Zihan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fang, Zihan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Wang, Qianru","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Qianru","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"An, Haonan","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"An, Haonan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Lin, Zheng","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lin, Zheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Deng, Yiqin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Deng, Yiqin","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":null,"display_name":"Chen, Xianhao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chen, Xianhao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":null,"display_name":"Fang, Yuguang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Fang, Yuguang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.41830000281333923,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.41830000281333923,"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/T11704","display_name":"Mobile Crowdsensing and Crowdsourcing","score":0.08780000358819962,"subfield":{"id":"https://openalex.org/subfields/1706","display_name":"Computer Science Applications"},"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.06440000236034393,"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/consistency","display_name":"Consistency (knowledge bases)","score":0.6531999707221985},{"id":"https://openalex.org/keywords/discriminative-model","display_name":"Discriminative model","score":0.5824000239372253},{"id":"https://openalex.org/keywords/semantic-heterogeneity","display_name":"Semantic heterogeneity","score":0.545199990272522},{"id":"https://openalex.org/keywords/weighting","display_name":"Weighting","score":0.5091000199317932},{"id":"https://openalex.org/keywords/data-aggregator","display_name":"Data aggregator","score":0.4918999969959259},{"id":"https://openalex.org/keywords/gating","display_name":"Gating","score":0.40689998865127563},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.40369999408721924},{"id":"https://openalex.org/keywords/stability","display_name":"Stability (learning theory)","score":0.39579999446868896}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8062000274658203},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.6531999707221985},{"id":"https://openalex.org/C97931131","wikidata":"https://www.wikidata.org/wiki/Q5282087","display_name":"Discriminative model","level":2,"score":0.5824000239372253},{"id":"https://openalex.org/C2778180026","wikidata":"https://www.wikidata.org/wiki/Q18378163","display_name":"Semantic heterogeneity","level":4,"score":0.545199990272522},{"id":"https://openalex.org/C183115368","wikidata":"https://www.wikidata.org/wiki/Q856577","display_name":"Weighting","level":2,"score":0.5091000199317932},{"id":"https://openalex.org/C82578977","wikidata":"https://www.wikidata.org/wiki/Q16773055","display_name":"Data aggregator","level":3,"score":0.4918999969959259},{"id":"https://openalex.org/C194544171","wikidata":"https://www.wikidata.org/wiki/Q21105679","display_name":"Gating","level":2,"score":0.40689998865127563},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.40369999408721924},{"id":"https://openalex.org/C112972136","wikidata":"https://www.wikidata.org/wiki/Q7595718","display_name":"Stability (learning theory)","level":2,"score":0.39579999446868896},{"id":"https://openalex.org/C93361087","wikidata":"https://www.wikidata.org/wiki/Q4426698","display_name":"Data consistency","level":2,"score":0.382999986410141},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.36980000138282776},{"id":"https://openalex.org/C74172769","wikidata":"https://www.wikidata.org/wiki/Q1446839","display_name":"Routing (electronic design automation)","level":2,"score":0.36739999055862427},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.3610999882221222},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3596999943256378},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.3467000126838684},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3391000032424927},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.3366999924182892},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.3278000056743622},{"id":"https://openalex.org/C19443361","wikidata":"https://www.wikidata.org/wiki/Q5282533","display_name":"Disparate system","level":2,"score":0.314300000667572},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.2944999933242798},{"id":"https://openalex.org/C72634772","wikidata":"https://www.wikidata.org/wiki/Q386824","display_name":"Data integration","level":2,"score":0.27970001101493835},{"id":"https://openalex.org/C59404180","wikidata":"https://www.wikidata.org/wiki/Q17013334","display_name":"Feature learning","level":2,"score":0.27950000762939453},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.26820001006126404},{"id":"https://openalex.org/C23123220","wikidata":"https://www.wikidata.org/wiki/Q816826","display_name":"Information retrieval","level":1,"score":0.2655999958515167},{"id":"https://openalex.org/C90312973","wikidata":"https://www.wikidata.org/wiki/Q7449052","display_name":"Semantic data model","level":2,"score":0.26109999418258667},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.25440001487731934}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.21276","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.21276","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2603.21276","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.21276","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/10","display_name":"Reduced inequalities","score":0.7097168564796448}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1],"models":[2],"(LLMs)":[3],"increasingly":[4],"adopt":[5],"Mixture-of-Experts":[6],"(MoE)":[7],"architectures":[8],"to":[9,23,39,47,70,83,96,114],"scale":[10],"model":[11],"capacity":[12],"while":[13],"reducing":[14],"computation.":[15],"Fine-tuning":[16],"these":[17,126],"MoE-based":[18,42,59],"LLMs":[19],"often":[20],"requires":[21],"access":[22],"distributed":[24],"and":[25,103,118,141,157,184,194,212],"privacy-sensitive":[26],"data,":[27],"making":[28],"centralized":[29],"fine-tuning":[30,61],"impractical.":[31],"Federated":[32],"learning":[33],"(FL)":[34],"therefore":[35],"provides":[36],"a":[37,98,131],"paradigm":[38],"collaboratively":[40],"fine-tune":[41],"LLMs,":[43],"enabling":[44],"each":[45],"client":[46],"integrate":[48],"diverse":[49],"knowledge":[50],"without":[51,168],"compromising":[52],"data":[53,72,79],"privacy.":[54],"However,":[55],"the":[56,119],"integration":[57],"of":[58,121],"LLM":[60],"into":[62],"FL":[63],"encounters":[64],"two":[65],"critical":[66],"aggregation":[67,95,133],"challenges":[68],"due":[69],"inherent":[71],"heterogeneity":[73],"across":[74,111,182],"clients:":[75],"(i)":[76],"divergent":[77],"local":[78,159,171],"distributions":[80,153],"drive":[81],"clients":[82,183],"develop":[84,107],"distinct":[85],"gating":[86,101,148,160],"preference":[87],"for":[88,198],"localized":[89],"expert":[90,115,122,142],"selection,":[91],"causing":[92],"direct":[93],"parameter":[94],"produce":[97],"``one-size-fits-none''":[99],"global":[100,199],"network,":[102],"(ii)":[104],"same-indexed":[105,180],"experts":[106,181],"disparate":[108],"semantic":[109,116,143,177],"roles":[110,197],"clients,":[112,191],"leading":[113],"blurring":[117],"degradation":[120],"specialization.":[123],"To":[124],"address":[125],"challenges,":[127],"we":[128],"propose":[129],"FedAlign-MoE,":[130],"federated":[132,217],"alignment":[134],"framework":[135],"that":[136,204],"jointly":[137],"enforces":[138],"routing":[139,152],"consistency":[140,155,178],"alignment.":[144],"Specifically,":[145],"FedAlign-MoE":[146,174,205],"aggregates":[147,186],"behaviors":[149],"by":[150],"aligning":[151],"through":[154,162],"weighting":[156],"optimizes":[158],"networks":[161],"distribution":[163],"regularization,":[164],"maintaining":[165],"cross-client":[166],"stability":[167],"overriding":[169],"discriminative":[170],"preferences.":[172],"Meanwhile,":[173],"explicitly":[175],"quantifies":[176],"among":[179],"selectively":[185],"updates":[187],"from":[188],"semantically":[189],"aligned":[190],"ensuring":[192],"stable":[193],"specialized":[195],"functional":[196],"experts.":[200],"Extensive":[201],"experiments":[202],"demonstrate":[203],"outperforms":[206],"state-of-the-art":[207],"benchmarks,":[208],"achieving":[209],"faster":[210],"convergence":[211],"superior":[213],"accuracy":[214],"in":[215],"non-IID":[216],"environments.":[218]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-03-25T00:00:00"}