{"id":"https://openalex.org/W7119085443","doi":"https://doi.org/10.48550/arxiv.2601.02967","title":"MoE Adapter for Large Audio Language Models: Sparsity, Disentanglement, and Gradient-Conflict-Free","display_name":"MoE Adapter for Large Audio Language Models: Sparsity, Disentanglement, and Gradient-Conflict-Free","publication_year":2026,"publication_date":"2026-01-06","ids":{"openalex":"https://openalex.org/W7119085443","doi":"https://doi.org/10.48550/arxiv.2601.02967"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2601.02967","doi":"https://doi.org/10.48550/arxiv.2601.02967","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2601.02967","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,"listed_in":[],"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"},"volume":null,"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2601.02967","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5046733996","display_name":"Yishu Lei","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Lei, Yishu","raw_affiliation_strings":["ERNIE Team , Baidu"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ERNIE Team , Baidu","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122280225","display_name":"Shuwei He","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I2722730","display_name":"Inner Mongolia University","ror":"https://ror.org/0106qb496","country_code":"CN","type":"education","lineage":["https://openalex.org/I2722730"]},{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"He, Shuwei","raw_affiliation_strings":["College of Computer Science , Inner Mongolia University","ERNIE Team , Baidu"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"College of Computer Science , Inner Mongolia University","institution_ids":["https://openalex.org/I2722730"]},{"raw_affiliation_string":"ERNIE Team , Baidu","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122300629","display_name":"Jing Hu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I4405263052","display_name":"Tsinghua Shenzhen International Graduate School","ror":"https://ror.org/01vceef04","country_code":"CN","type":"education","lineage":["https://openalex.org/I4405263052","https://openalex.org/I99065089"]},{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]},{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hu, Jing","raw_affiliation_strings":["ERNIE Team , Baidu","Tsinghua Shenzhen International Graduate School , Tsinghua University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ERNIE Team , Baidu","institution_ids":["https://openalex.org/I98301712"]},{"raw_affiliation_string":"Tsinghua Shenzhen International Graduate School , Tsinghua University","institution_ids":["https://openalex.org/I4405263052","https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Zhang, Dan","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhang, Dan","raw_affiliation_strings":["ERNIE Team , Baidu"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ERNIE Team , Baidu","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102611212","display_name":"Xianlong Luo","orcid":"https://orcid.org/0009-0008-9607-232X","observed_orcids":["https://orcid.org/0009-0008-9607-232X"]},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Luo, Xianlong","raw_affiliation_strings":["ERNIE Team , Baidu"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ERNIE Team , Baidu","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5041664963","display_name":"Danxiang Zhu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhu, Danxiang","raw_affiliation_strings":["ERNIE Team , Baidu"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ERNIE Team , Baidu","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5095102135","display_name":"Shikun Feng","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Feng, Shikun","raw_affiliation_strings":["ERNIE Team , Baidu"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ERNIE Team , Baidu","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137132732","display_name":"Rui Liu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I2722730","display_name":"Inner Mongolia University","ror":"https://ror.org/0106qb496","country_code":"CN","type":"education","lineage":["https://openalex.org/I2722730"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Liu, Rui","raw_affiliation_strings":["College of Computer Science , Inner Mongolia University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"College of Computer Science , Inner Mongolia University","institution_ids":["https://openalex.org/I2722730"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122288228","display_name":"Jingzhou He","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"He, Jingzhou","raw_affiliation_strings":["ERNIE Team , Baidu"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ERNIE Team , Baidu","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122232434","display_name":"Yu Sun","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Sun, Yu","raw_affiliation_strings":["ERNIE Team , Baidu"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ERNIE Team , Baidu","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5140927288","display_name":"Hua Wu","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wu, Hua","raw_affiliation_strings":["ERNIE Team , Baidu"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ERNIE Team , Baidu","institution_ids":["https://openalex.org/I98301712"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5125251827","display_name":"Haifeng Wang","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I98301712","display_name":"Baidu (China)","ror":"https://ror.org/03vs3wt56","country_code":"CN","type":"company","lineage":["https://openalex.org/I98301712"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wang, Haifeng","raw_affiliation_strings":["ERNIE Team , Baidu"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ERNIE Team , Baidu","institution_ids":["https://openalex.org/I98301712"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.9369999766349792,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.9369999766349792,"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/T11309","display_name":"Music and Audio Processing","score":0.02969999983906746,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T10860","display_name":"Speech and Audio Processing","score":0.01679999940097332,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/mixture-of-experts","display_name":"mixture of experts","score":0.800000011920929},{"id":"https://openalex.org/keywords/audio-language-models","display_name":"audio-language models","score":0.8740000128746033},{"id":"https://openalex.org/keywords/gradient-conflict","display_name":"gradient conflict","score":0.746999979019165},{"id":"https://openalex.org/keywords/dynamic-gating","display_name":"dynamic gating","score":0.6050000190734863},{"id":"https://openalex.org/keywords/sparsity","display_name":"sparsity","score":0.6069999933242798},{"id":"https://openalex.org/keywords/disentanglement","display_name":"disentanglement","score":0.765999972820282},{"id":"https://openalex.org/keywords/multimodal-large-language-models","display_name":"multimodal large language models","score":0.7419999837875366},{"id":"https://openalex.org/keywords/paralanguage","display_name":"paralinguistic features","score":0.8500000238418579}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8241999745368958},{"id":"https://openalex.org/C177284502","wikidata":"https://www.wikidata.org/wiki/Q1005390","display_name":"Adapter (computing)","level":2,"score":0.6444000005722046},{"id":"https://openalex.org/C133378560","wikidata":"https://www.wikidata.org/wiki/Q1753225","display_name":"Paralanguage","level":2,"score":0.5062000155448914},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.4627000093460083},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.45750001072883606},{"id":"https://openalex.org/C36503486","wikidata":"https://www.wikidata.org/wiki/Q11235244","display_name":"Domain (mathematical analysis)","level":2,"score":0.4090000092983246},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.37790000438690186},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.37470000982284546},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.35339999198913574},{"id":"https://openalex.org/C12362212","wikidata":"https://www.wikidata.org/wiki/Q728435","display_name":"Linear subspace","level":2,"score":0.35100001096725464},{"id":"https://openalex.org/C2780226545","wikidata":"https://www.wikidata.org/wiki/Q6888030","display_name":"Modality (human\u2013computer interaction)","level":2,"score":0.34630000591278076},{"id":"https://openalex.org/C2778263558","wikidata":"https://www.wikidata.org/wiki/Q46384","display_name":"Microphone","level":3,"score":0.34549999237060547},{"id":"https://openalex.org/C157657479","wikidata":"https://www.wikidata.org/wiki/Q2367247","display_name":"Closed captioning","level":3,"score":0.33090001344680786},{"id":"https://openalex.org/C64922751","wikidata":"https://www.wikidata.org/wiki/Q4650799","display_name":"Audio signal","level":3,"score":0.323199987411499},{"id":"https://openalex.org/C107457646","wikidata":"https://www.wikidata.org/wiki/Q207434","display_name":"Human\u2013computer interaction","level":1,"score":0.32120001316070557},{"id":"https://openalex.org/C522192633","wikidata":"https://www.wikidata.org/wiki/Q34228","display_name":"Sign language","level":2,"score":0.2994000017642975},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.2815000116825104},{"id":"https://openalex.org/C194544171","wikidata":"https://www.wikidata.org/wiki/Q21105679","display_name":"Gating","level":2,"score":0.28040000796318054},{"id":"https://openalex.org/C124304363","wikidata":"https://www.wikidata.org/wiki/Q673661","display_name":"Abstraction","level":2,"score":0.2759000062942505},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.2700999975204468},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.26899999380111694},{"id":"https://openalex.org/C2776230583","wikidata":"https://www.wikidata.org/wiki/Q1322198","display_name":"Spoken language","level":2,"score":0.2556999921798706},{"id":"https://openalex.org/C135641252","wikidata":"https://www.wikidata.org/wiki/Q738567","display_name":"Multimodal interaction","level":2,"score":0.250900000333786}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2601.02967","doi":"https://doi.org/10.48550/arxiv.2601.02967","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2601.02967","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,"listed_in":[],"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.2601.02967","doi":"https://doi.org/10.48550/arxiv.2601.02967","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2601.02967","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,"listed_in":[],"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":[],"sustainable_development_goals_aurora":[{"score":0.5023664236068726,"id":"https://metadata.un.org/sdg/4","display_name":"Quality education"}],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Extending":[0],"the":[1,9,74,123,148],"input":[2],"modality":[3],"of":[4],"Large":[5],"Language":[6],"Models~(LLMs)":[7],"to":[8,42,47,81,96,153],"audio":[10,94,130],"domain":[11],"is":[12,21,26,40],"essential":[13],"for":[14,62,107],"achieving":[15],"comprehensive":[16],"multimodal":[17],"perception.":[18],"However,":[19],"it":[20,86],"well-known":[22],"that":[23,92,122],"acoustic":[24,83],"information":[25],"intrinsically":[27],"\\textit{heterogeneous},":[28],"entangling":[29],"attributes":[30,64],"such":[31],"as":[32,58],"speech,":[33],"music,":[34],"and":[35,114,132,151],"environmental":[36],"context.":[37],"Existing":[38],"research":[39],"limited":[41],"a":[43,76,88],"dense,":[44],"parameter-shared":[45],"adapter":[46],"model":[48],"these":[49],"diverse":[50],"patterns,":[51],"which":[52],"induces":[53],"\\textit{gradient":[54],"conflict}":[55],"during":[56],"optimization,":[57],"parameter":[59],"updates":[60],"required":[61],"distinct":[63],"contradict":[65],"each":[66],"other.":[67],"To":[68],"address":[69],"this":[70],"limitation,":[71],"we":[72,145],"introduce":[73],"\\textit{\\textbf{MoE-Adapter}},":[75],"sparse":[77],"Mixture-of-Experts~(MoE)":[78],"architecture":[79],"designed":[80],"decouple":[82],"information.":[84],"Specifically,":[85],"employs":[87],"dynamic":[89],"gating":[90],"mechanism":[91],"routes":[93],"tokens":[95],"specialized":[97],"experts":[98,106],"capturing":[99],"complementary":[100],"feature":[101,117],"subspaces":[102],"while":[103],"retaining":[104],"shared":[105],"global":[108],"context,":[109],"thereby":[110],"mitigating":[111],"gradient":[112],"conflicts":[113],"enabling":[115],"fine-grained":[116],"learning.":[118],"Comprehensive":[119],"experiments":[120],"show":[121],"MoE-Adapter":[124],"achieves":[125],"superior":[126],"performance":[127],"on":[128],"both":[129],"semantic":[131],"paralinguistic":[133],"tasks,":[134],"consistently":[135],"outperforming":[136],"dense":[137],"linear":[138],"baselines":[139],"with":[140],"comparable":[141],"computational":[142],"costs.":[143],"Furthermore,":[144],"will":[146],"release":[147],"related":[149],"code":[150],"models":[152],"facilitate":[154],"future":[155],"research.":[156]},"counts_by_year":[],"updated_date":"2026-10-14T16:39:01.820549","created_date":"2026-01-08T00:00:00"}