{"id":"https://openalex.org/W4413145346","doi":"https://doi.org/10.1109/cvpr52734.2025.00943","title":"Self-Expansion of Pre-trained Models with Mixture of Adapters for Continual Learning","display_name":"Self-Expansion of Pre-trained Models with Mixture of Adapters for Continual Learning","publication_year":2025,"publication_date":"2025-06-10","ids":{"openalex":"https://openalex.org/W4413145346","doi":"https://doi.org/10.1109/cvpr52734.2025.00943"},"language":"en","primary_location":{"id":"doi:10.1109/cvpr52734.2025.00943","is_oa":false,"landing_page_url":"https://doi.org/10.1109/cvpr52734.2025.00943","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","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/A5052104663","display_name":"Huiyi Wang","orcid":"https://orcid.org/0009-0005-7769-8317"},"institutions":[{"id":"https://openalex.org/I31746571","display_name":"UNSW Sydney","ror":"https://ror.org/03r8z3t63","country_code":"AU","type":"education","lineage":["https://openalex.org/I31746571"]}],"countries":["AU"],"is_corresponding":false,"raw_author_name":"Huiyi Wang","raw_affiliation_strings":["University of New South Wales"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of New South Wales","institution_ids":["https://openalex.org/I31746571"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101722089","display_name":"Haodong Lu","orcid":"https://orcid.org/0000-0002-8428-7023"},"institutions":[{"id":"https://openalex.org/I31746571","display_name":"UNSW Sydney","ror":"https://ror.org/03r8z3t63","country_code":"AU","type":"education","lineage":["https://openalex.org/I31746571"]}],"countries":["AU"],"is_corresponding":false,"raw_author_name":"Haodong Lu","raw_affiliation_strings":["University of New South Wales"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of New South Wales","institution_ids":["https://openalex.org/I31746571"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103021890","display_name":"Lina Yao","orcid":"https://orcid.org/0000-0003-2772-0385"},"institutions":[{"id":"https://openalex.org/I1292875679","display_name":"Commonwealth Scientific and Industrial Research Organisation","ror":"https://ror.org/03qn8fb07","country_code":"AU","type":"government","lineage":["https://openalex.org/I1292875679","https://openalex.org/I2801453606","https://openalex.org/I4387156119"]},{"id":"https://openalex.org/I42894916","display_name":"Data61","ror":"https://ror.org/03q397159","country_code":"AU","type":"other","lineage":["https://openalex.org/I1292875679","https://openalex.org/I2801453606","https://openalex.org/I42894916","https://openalex.org/I4387156119"]}],"countries":["AU"],"is_corresponding":false,"raw_author_name":"Lina Yao","raw_affiliation_strings":["CSIRO’s Data61"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"CSIRO’s Data61","institution_ids":["https://openalex.org/I1292875679","https://openalex.org/I42894916"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5101768095","display_name":"Dong Gong","orcid":"https://orcid.org/0000-0002-2668-9630"},"institutions":[{"id":"https://openalex.org/I31746571","display_name":"UNSW Sydney","ror":"https://ror.org/03r8z3t63","country_code":"AU","type":"education","lineage":["https://openalex.org/I31746571"]}],"countries":["AU"],"is_corresponding":false,"raw_author_name":"Dong Gong","raw_affiliation_strings":["University of New South Wales"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of New South Wales","institution_ids":["https://openalex.org/I31746571"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":11,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"10087","last_page":"10098"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.9710999727249146,"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":0.9710999727249146,"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":0.6418278217315674},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.45193397998809814}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6418278217315674},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.45193397998809814}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/cvpr52734.2025.00943","is_oa":false,"landing_page_url":"https://doi.org/10.1109/cvpr52734.2025.00943","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2025 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)","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":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"Continual":[0],"learning":[1],"(CL)":[2],"aims":[3],"to":[4,71,106,119,170,214],"continually":[5],"accumulate":[6],"knowledge":[7,91,194],"from":[8,75],"a":[9,20,65,103,151,155,164],"non-stationary":[10],"data":[11],"stream":[12],"without":[13,218],"catastrophic":[14],"forgetting":[15],"of":[16,68,96,110,150,188,205],"learned":[17,184],"knowledge,":[18],"requiring":[19],"balance":[21,112],"between":[22],"stability":[23],"and":[24,89,154,168,196],"adaptability.":[25],"Relying":[26],"on":[27,42,64,125,130],"the":[28,52,108,177,203,206],"generalizable":[29],"representation":[30,143,156,159],"in":[31,84,113,127],"pre-trained":[32,97],"models":[33,98],"(PTMs),":[34],"PTM-based":[35,58,114,215],"CL":[36,59,77,216],"methods":[37,60,217],"perform":[38],"effective":[39],"continual":[40],"adaptation":[41,63],"downstream":[43],"tasks":[44],"by":[45],"adding":[46,80],"learnable":[47],"adapters":[48],"or":[49,121],"prompts":[50],"upon":[51],"frozen":[53],"PTMs.":[54],"However,":[55],"many":[56],"existing":[57],"use":[61],"restricted":[62],"fixed":[66],"set":[67],"these":[69],"modules":[70,82,124],"avoid":[72],"forgetting,":[73],"suffering":[74],"limited":[76],"ability.":[78],"Periodically":[79],"task-specific":[81],"results":[83],"linear":[85],"model":[86],"growth":[87],"rate":[88],"impaired":[90],"reuse.":[92],"We":[93,145],"propose":[94],"Self-Expansion":[95],"with":[99],"Modularized":[100],"Adaptation":[101],"(SEMA),":[102],"novel":[104],"approach":[105],"enhance":[107],"control":[109],"stability-plasticity":[111],"CL.":[115],"SEMA":[116,191],"automatically":[117],"decides":[118],"reuse":[120,195],"add":[122],"adapter":[123,148,153,189],"demand":[126],"CL,":[128],"depending":[129],"whether":[131],"significant":[132],"distribution":[133,165],"shift":[134,166],"that":[135],"cannot":[136],"be":[137],"handled":[138],"is":[139,183,222],"detected":[140],"at":[141,224],"different":[142],"levels.":[144],"design":[146],"modular":[147],"consisting":[149],"functional":[152],"descriptor.":[157],"The":[158],"descriptors":[160],"are":[161],"trained":[162],"as":[163],"indicator":[167],"used":[169],"trigger":[171],"self-expansion":[172,208],"signals.":[173],"For":[174],"better":[175,193],"composing":[176],"adapters,":[178],"an":[179],"expandable":[180],"weighting":[181],"router":[182],"jointly":[185],"for":[186],"mixture":[187],"outputs.":[190],"enables":[192],"sub-linear":[197],"expansion":[198],"rate.":[199],"Extensive":[200],"experiments":[201],"demonstrate":[202],"effectiveness":[204],"proposed":[207],"method,":[209],"achieving":[210],"state-of-the-art":[211],"performance":[212],"compared":[213],"memory":[219],"rehearsal.":[220],"Code":[221],"available":[223],"https://github.com/huiyiwang01/SEMA-CL.":[225]},"counts_by_year":[{"year":2026,"cited_by_count":7},{"year":2025,"cited_by_count":4}],"updated_date":"2026-07-22T07:51:19.307946","created_date":"2025-10-10T00:00:00"}