{"id":"https://openalex.org/W4415539583","doi":"https://doi.org/10.1145/3746027.3755586","title":"Multi-Task Dense Prediction Fine-Tuning with Mixture of Fine-Grained Experts","display_name":"Multi-Task Dense Prediction Fine-Tuning with Mixture of Fine-Grained Experts","publication_year":2025,"publication_date":"2025-10-25","ids":{"openalex":"https://openalex.org/W4415539583","doi":"https://doi.org/10.1145/3746027.3755586"},"language":"en","primary_location":{"id":"doi:10.1145/3746027.3755586","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3746027.3755586","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd ACM International Conference on Multimedia","raw_type":"proceedings-article"},"type":"preprint","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3746027.3755586","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5077429843","display_name":"Yangyang Xu","orcid":"https://orcid.org/0000-0002-7098-8419"},"institutions":[{"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":true,"raw_author_name":"Yangyang Xu","raw_affiliation_strings":["Department of Computer Science and Technology, Tsinghua University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-7098-8419","affiliations":[{"raw_affiliation_string":"Department of Computer Science and Technology, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Xi Ye","orcid":"https://orcid.org/0000-0002-1763-6019"},"institutions":[{"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":"Xi Ye","raw_affiliation_strings":["Department of Computer Science and Technology, Tsinghua University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-1763-6019","affiliations":[{"raw_affiliation_string":"Department of Computer Science and Technology, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5087092119","display_name":"Duo Su","orcid":"https://orcid.org/0000-0002-9607-3639"},"institutions":[{"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":"Duo Su","raw_affiliation_strings":["Department of Computer Science and Technology, Tsinghua University, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-9607-3639","affiliations":[{"raw_affiliation_string":"Department of Computer Science and Technology, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5077429843"],"corresponding_institution_ids":["https://openalex.org/I99065089"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.15612103,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"4758","last_page":"4767"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10320","display_name":"Neural Networks and Applications","score":0.9958000183105469,"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/T10320","display_name":"Neural Networks and Applications","score":0.9958000183105469,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9957000017166138,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.9954000115394592,"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/key","display_name":"Key (lock)","score":0.635200023651123},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.5995000004768372},{"id":"https://openalex.org/keywords/focus","display_name":"Focus (optics)","score":0.5306000113487244},{"id":"https://openalex.org/keywords/feature","display_name":"Feature (linguistics)","score":0.5034999847412109},{"id":"https://openalex.org/keywords/partition","display_name":"Partition (number theory)","score":0.43650001287460327},{"id":"https://openalex.org/keywords/transfer-of-learning","display_name":"Transfer of learning","score":0.38580000400543213},{"id":"https://openalex.org/keywords/feature-learning","display_name":"Feature learning","score":0.35359999537467957},{"id":"https://openalex.org/keywords/task-analysis","display_name":"Task analysis","score":0.3508000075817108}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7515000104904175},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.635200023651123},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.5995000004768372},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5861999988555908},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5723999738693237},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.5306000113487244},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.5034999847412109},{"id":"https://openalex.org/C42812","wikidata":"https://www.wikidata.org/wiki/Q1082910","display_name":"Partition (number theory)","level":2,"score":0.43650001287460327},{"id":"https://openalex.org/C150899416","wikidata":"https://www.wikidata.org/wiki/Q1820378","display_name":"Transfer of learning","level":2,"score":0.38580000400543213},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.38119998574256897},{"id":"https://openalex.org/C59404180","wikidata":"https://www.wikidata.org/wiki/Q17013334","display_name":"Feature learning","level":2,"score":0.35359999537467957},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.3508000075817108},{"id":"https://openalex.org/C22367795","wikidata":"https://www.wikidata.org/wiki/Q7625208","display_name":"Structured prediction","level":2,"score":0.33320000767707825},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.3273000121116638},{"id":"https://openalex.org/C124681953","wikidata":"https://www.wikidata.org/wiki/Q339062","display_name":"Decomposition","level":2,"score":0.31859999895095825},{"id":"https://openalex.org/C2776604539","wikidata":"https://www.wikidata.org/wiki/Q6423395","display_name":"Knowledge sharing","level":2,"score":0.3122999966144562},{"id":"https://openalex.org/C2776854237","wikidata":"https://www.wikidata.org/wiki/Q6031064","display_name":"Information sharing","level":2,"score":0.3003999888896942},{"id":"https://openalex.org/C2776960227","wikidata":"https://www.wikidata.org/wiki/Q2586354","display_name":"Knowledge transfer","level":2,"score":0.2874000072479248},{"id":"https://openalex.org/C45804977","wikidata":"https://www.wikidata.org/wiki/Q7239673","display_name":"Predictive modelling","level":2,"score":0.2757999897003174},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.2655999958515167},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.2558000087738037},{"id":"https://openalex.org/C2778334786","wikidata":"https://www.wikidata.org/wiki/Q1586270","display_name":"Variation (astronomy)","level":2,"score":0.2551000118255615},{"id":"https://openalex.org/C74172769","wikidata":"https://www.wikidata.org/wiki/Q1446839","display_name":"Routing (electronic design automation)","level":2,"score":0.2538999915122986}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1145/3746027.3755586","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3746027.3755586","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd ACM International Conference on Multimedia","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2507.19077","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2507.19077","pdf_url":"https://arxiv.org/pdf/2507.19077","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"doi:10.1145/3746027.3755586","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3746027.3755586","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd ACM International Conference on Multimedia","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":6,"referenced_works":["https://openalex.org/W2110898478","https://openalex.org/W3138516171","https://openalex.org/W4382240003","https://openalex.org/W4383503846","https://openalex.org/W4390590999","https://openalex.org/W4393160572"],"related_works":[],"abstract_inverted_index":{"Multi-task":[0],"learning":[1],"(MTL)":[2],"for":[3],"dense":[4,167],"prediction":[5,168],"has":[6],"shown":[7],"promising":[8],"results":[9,148],"but":[10],"still":[11],"faces":[12],"challenges":[13],"in":[14,173],"balancing":[15],"shared":[16,74],"representations":[17],"with":[18],"task-specific":[19],"specialization.":[20],"In":[21,127],"this":[22],"paper,":[23],"we":[24,49,72,99,129],"introduce":[25,73],"a":[26,40,101],"novel":[27],"Fine-Grained":[28],"Mixture":[29],"of":[30,42,59,64,83,143],"Experts":[31],"(FGMoE)":[32],"architecture":[33],"that":[34,53,76,104,150],"explores":[35],"MoE-based":[36,161],"MTL":[37,163],"models":[38,164],"through":[39],"combination":[41],"three":[43],"key":[44],"innovations":[45],"and":[46,89,116,157],"fine-tuning.":[47],"First,":[48],"propose":[50],"intra-task":[51],"experts":[52,75,92],"partition":[54],"along":[55],"intermediate":[56],"hidden":[57],"dimensions":[58],"MLPs,":[60],"enabling":[61],"finer":[62],"decomposition":[63],"task":[65,117],"information":[66,79,121],"while":[67,123],"maintaining":[68],"parameter":[69,136],"efficiency.":[70],"Second,":[71],"consolidate":[77],"common":[78],"across":[80,109],"different":[81],"contexts":[82],"the":[84,131,141,144,151],"same":[85],"task,":[86],"reducing":[87],"redundancy,":[88],"allowing":[90],"routing":[91],"to":[93,134],"focus":[94],"on":[95,112,165],"unique":[96],"aspects.":[97],"Third,":[98],"design":[100],"global":[102],"expert":[103],"facilitates":[105],"adaptive":[106],"knowledge":[107],"transfer":[108],"tasks":[110],"based":[111],"both":[113],"input":[114],"feature":[115],"requirements,":[118],"promoting":[119],"beneficial":[120],"sharing":[122],"preventing":[124],"harmful":[125],"interference.":[126],"addition,":[128],"use":[130],"fine-tuning":[132],"approach":[133],"improve":[135],"efficiency":[137],"only":[138],"by":[139],"training":[140],"parameters":[142,156],"decoder.":[145],"Extensive":[146],"experimental":[147],"show":[149],"proposed":[152],"FGMoE":[153],"uses":[154],"fewer":[155],"significantly":[158],"outperforms":[159],"current":[160],"competitive":[162],"two":[166],"datasets":[169],"(i.e.,":[170],"NYUD-v2,":[171],"PASCAL-Context)":[172],"various":[174],"metrics.":[175]},"counts_by_year":[],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-25T00:00:00"}