{"id":"https://openalex.org/W4372259766","doi":"https://doi.org/10.1109/icassp49357.2023.10094792","title":"Boosting Bert Subnets with Neural Grafting","display_name":"Boosting Bert Subnets with Neural Grafting","publication_year":2023,"publication_date":"2023-05-05","ids":{"openalex":"https://openalex.org/W4372259766","doi":"https://doi.org/10.1109/icassp49357.2023.10094792"},"language":"en","primary_location":{"id":"doi:10.1109/icassp49357.2023.10094792","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/icassp49357.2023.10094792","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","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/A5101579165","display_name":"Ting Hu","orcid":"https://orcid.org/0000-0001-6266-2622"},"institutions":[{"id":"https://openalex.org/I143288331","display_name":"Hasso Plattner Institute","ror":"https://ror.org/058rn5r42","country_code":"DE","type":"facility","lineage":["https://openalex.org/I143288331","https://openalex.org/I176453806"]},{"id":"https://openalex.org/I176453806","display_name":"University of Potsdam","ror":"https://ror.org/03bnmw459","country_code":"DE","type":"education","lineage":["https://openalex.org/I176453806"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Ting Hu","raw_affiliation_strings":["University of Potsdam,Hasso Plattner Institute,Potsdam,Germany","Hasso Plattner Institute, University of Potsdam, Potsdam, Germany"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Potsdam,Hasso Plattner Institute,Potsdam,Germany","institution_ids":["https://openalex.org/I143288331","https://openalex.org/I176453806"]},{"raw_affiliation_string":"Hasso Plattner Institute, University of Potsdam, Potsdam, Germany","institution_ids":["https://openalex.org/I143288331","https://openalex.org/I176453806"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102919398","display_name":"Christoph Meinel","orcid":"https://orcid.org/0000-0002-3410-3193"},"institutions":[{"id":"https://openalex.org/I143288331","display_name":"Hasso Plattner Institute","ror":"https://ror.org/058rn5r42","country_code":"DE","type":"facility","lineage":["https://openalex.org/I143288331","https://openalex.org/I176453806"]},{"id":"https://openalex.org/I176453806","display_name":"University of Potsdam","ror":"https://ror.org/03bnmw459","country_code":"DE","type":"education","lineage":["https://openalex.org/I176453806"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Christoph Meinel","raw_affiliation_strings":["University of Potsdam,Hasso Plattner Institute,Potsdam,Germany","Hasso Plattner Institute, University of Potsdam, Potsdam, Germany"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Potsdam,Hasso Plattner Institute,Potsdam,Germany","institution_ids":["https://openalex.org/I143288331","https://openalex.org/I176453806"]},{"raw_affiliation_string":"Hasso Plattner Institute, University of Potsdam, Potsdam, Germany","institution_ids":["https://openalex.org/I143288331","https://openalex.org/I176453806"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5101039243","display_name":"Haojin Yang","orcid":null},"institutions":[{"id":"https://openalex.org/I143288331","display_name":"Hasso Plattner Institute","ror":"https://ror.org/058rn5r42","country_code":"DE","type":"facility","lineage":["https://openalex.org/I143288331","https://openalex.org/I176453806"]},{"id":"https://openalex.org/I176453806","display_name":"University of Potsdam","ror":"https://ror.org/03bnmw459","country_code":"DE","type":"education","lineage":["https://openalex.org/I176453806"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Haojin Yang","raw_affiliation_strings":["University of Potsdam,Hasso Plattner Institute,Potsdam,Germany","Hasso Plattner Institute, University of Potsdam, Potsdam, Germany"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Potsdam,Hasso Plattner Institute,Potsdam,Germany","institution_ids":["https://openalex.org/I143288331","https://openalex.org/I176453806"]},{"raw_affiliation_string":"Hasso Plattner Institute, University of Potsdam, Potsdam, Germany","institution_ids":["https://openalex.org/I143288331","https://openalex.org/I176453806"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.03438818,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"33","issue":null,"first_page":"1","last_page":"5"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9998000264167786,"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/T10028","display_name":"Topic Modeling","score":0.9998000264167786,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9994999766349792,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.9929999709129333,"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":0.8292988538742065},{"id":"https://openalex.org/keywords/subnet","display_name":"Subnet","score":0.7887848615646362},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.5647528767585754},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.5492253303527832},{"id":"https://openalex.org/keywords/boosting","display_name":"Boosting (machine learning)","score":0.5373368263244629},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.5063761472702026},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.4690554440021515},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.36047035455703735},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.22572433948516846},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.20086395740509033}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8292988538742065},{"id":"https://openalex.org/C21099817","wikidata":"https://www.wikidata.org/wiki/Q7631721","display_name":"Subnet","level":2,"score":0.7887848615646362},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.5647528767585754},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.5492253303527832},{"id":"https://openalex.org/C46686674","wikidata":"https://www.wikidata.org/wiki/Q466303","display_name":"Boosting (machine learning)","level":2,"score":0.5373368263244629},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.5063761472702026},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4690554440021515},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.36047035455703735},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.22572433948516846},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.20086395740509033},{"id":"https://openalex.org/C13280743","wikidata":"https://www.wikidata.org/wiki/Q131089","display_name":"Geodesy","level":1,"score":0.0},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.0},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp49357.2023.10094792","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/icassp49357.2023.10094792","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","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":22,"referenced_works":["https://openalex.org/W2510886230","https://openalex.org/W2896457183","https://openalex.org/W2923014074","https://openalex.org/W2970454332","https://openalex.org/W2975381464","https://openalex.org/W2978017171","https://openalex.org/W2994749257","https://openalex.org/W3015609966","https://openalex.org/W3105966348","https://openalex.org/W3112673818","https://openalex.org/W3160843329","https://openalex.org/W3167851353","https://openalex.org/W4287021475","https://openalex.org/W4287391717","https://openalex.org/W4288089799","https://openalex.org/W4385245566","https://openalex.org/W6739901393","https://openalex.org/W6767064347","https://openalex.org/W6768080748","https://openalex.org/W6769627184","https://openalex.org/W6775706467","https://openalex.org/W6788811087"],"related_works":["https://openalex.org/W2102539527","https://openalex.org/W2131631951","https://openalex.org/W2356206668","https://openalex.org/W2130707537","https://openalex.org/W3200778902","https://openalex.org/W1977409556","https://openalex.org/W2361602549","https://openalex.org/W4362683600","https://openalex.org/W2105155969","https://openalex.org/W2390667963"],"abstract_inverted_index":{"Pre-trained":[0],"Language":[1,5],"Models":[2],"in":[3,24,54],"Natural":[4],"Processing":[6],"have":[7],"become":[8],"increasingly":[9],"computationally":[10],"expensive":[11],"and":[12,69,101,139,145],"memory":[13],"demanding.":[14],"The":[15,51,128],"recently":[16],"proposed":[17,113],"computation-adaptive":[18],"BERT":[19,30,81,96,120],"models":[20],"facilitate":[21],"their":[22,45],"deployment":[23],"practical":[25],"applications.":[26],"Training":[27],"such":[28],"a":[29,61,95],"model":[31,97],"involves":[32],"jointly":[33],"optimizing":[34],"subnets":[35,53,121],"of":[36,94,119,125,167],"varying":[37],"sizes,":[38],"which":[39],"is":[40,60],"not":[41],"easy":[42],"due":[43],"to":[44,79,132,160],"mutual":[46],"interference":[47],"with":[48],"one":[49],"another.":[50],"larger-size":[52],"particular":[55],"could":[56],"deteriorate":[57],"when":[58],"there":[59],"large":[62],"performance":[63,118],"gap":[64],"between":[65],"the":[66,70,84,90,112,116,133,164],"smallest":[67],"subnet":[68,129],"super-net.":[71],"In":[72],"this":[73],"work,":[74],"we":[75,88,149],"propose":[76],"Neural":[77,106,152,168],"grafting":[78,153],"boost":[80],"subnets,":[82],"especially":[83],"larger":[85],"ones.":[86],"Specifically,":[87],"regard":[89],"less":[91,99],"important":[92],"sub-modules":[93],"as":[98],"active":[100],"reactivate":[102],"them":[103],"via":[104],"layer-wise":[105],"grafting.":[107,169],"Experimental":[108],"results":[109],"show":[110],"that":[111],"method":[114],"improves":[115],"average":[117],"on":[122,143,163],"six":[123],"datasets":[124],"GLUE":[126],"benchmark.":[127],"performing":[130],"comparable":[131],"supernet":[134],"BERT-Base":[135],"reduces":[136],"around":[137],"67%":[138],"70%":[140],"inference":[141],"latency":[142],"GPU":[144],"CPU,":[146],"respectively.":[147],"Moreover,":[148],"compare":[150],"two":[151],"strategies":[154],"under":[155],"varied":[156],"experimental":[157],"settings,":[158],"hoping":[159],"shed":[161],"light":[162],"application":[165],"scenarios":[166]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}