{"id":"https://openalex.org/W3160106041","doi":"https://doi.org/10.1109/icassp39728.2021.9414572","title":"Elbert: Fast Albert with Confidence-Window Based Early Exit","display_name":"Elbert: Fast Albert with Confidence-Window Based Early Exit","publication_year":2021,"publication_date":"2021-05-13","ids":{"openalex":"https://openalex.org/W3160106041","doi":"https://doi.org/10.1109/icassp39728.2021.9414572","mag":"3160106041"},"language":"en","primary_location":{"id":"doi:10.1109/icassp39728.2021.9414572","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp39728.2021.9414572","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2021 - 2021 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/A5078409838","display_name":"Keli Xie","orcid":"https://orcid.org/0000-0002-0021-5285"},"institutions":[{"id":"https://openalex.org/I881766915","display_name":"Nanjing University","ror":"https://ror.org/01rxvg760","country_code":"CN","type":"education","lineage":["https://openalex.org/I881766915"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Keli Xie","raw_affiliation_strings":["School of Electronic Science and Engineering, Nanjing University, Nanjing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Electronic Science and Engineering, Nanjing University, Nanjing, China","institution_ids":["https://openalex.org/I881766915"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5084117570","display_name":"Siyuan Lu","orcid":"https://orcid.org/0000-0002-8639-3081"},"institutions":[{"id":"https://openalex.org/I881766915","display_name":"Nanjing University","ror":"https://ror.org/01rxvg760","country_code":"CN","type":"education","lineage":["https://openalex.org/I881766915"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Siyuan Lu","raw_affiliation_strings":["School of Electronic Science and Engineering, Nanjing University, Nanjing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Electronic Science and Engineering, Nanjing University, Nanjing, China","institution_ids":["https://openalex.org/I881766915"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100785858","display_name":"Meiqi Wang","orcid":"https://orcid.org/0000-0001-9553-3640"},"institutions":[{"id":"https://openalex.org/I881766915","display_name":"Nanjing University","ror":"https://ror.org/01rxvg760","country_code":"CN","type":"education","lineage":["https://openalex.org/I881766915"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Meiqi Wang","raw_affiliation_strings":["School of Electronic Science and Engineering, Nanjing University, Nanjing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Electronic Science and Engineering, Nanjing University, Nanjing, China","institution_ids":["https://openalex.org/I881766915"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100696999","display_name":"Zhongfeng Wang","orcid":"https://orcid.org/0000-0002-7227-4786"},"institutions":[{"id":"https://openalex.org/I881766915","display_name":"Nanjing University","ror":"https://ror.org/01rxvg760","country_code":"CN","type":"education","lineage":["https://openalex.org/I881766915"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhongfeng Wang","raw_affiliation_strings":["School of Electronic Science and Engineering, Nanjing University, Nanjing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Electronic Science and Engineering, Nanjing University, Nanjing, China","institution_ids":["https://openalex.org/I881766915"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I881766915"],"apc_list":null,"apc_paid":null,"fwci":2.0514,"has_fulltext":false,"cited_by_count":18,"citation_normalized_percentile":{"value":0.89231845,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"7713","last_page":"7717"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9998999834060669,"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":0.9998999834060669,"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":0.9994000196456909,"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":0.9905999898910522,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.8220977783203125},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.8107188940048218},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.8064577579498291},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.5774590373039246},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.5449042916297913},{"id":"https://openalex.org/keywords/window","display_name":"Window (computing)","score":0.5265567302703857},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.4836888909339905},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.472257524728775},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.448725163936615},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.4036848545074463},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.36116915941238403},{"id":"https://openalex.org/keywords/theoretical-computer-science","display_name":"Theoretical computer science","score":0.34182554483413696},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.17220449447631836},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.15436604619026184}],"concepts":[{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.8220977783203125},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.8107188940048218},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8064577579498291},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.5774590373039246},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5449042916297913},{"id":"https://openalex.org/C2778751112","wikidata":"https://www.wikidata.org/wiki/Q835016","display_name":"Window (computing)","level":2,"score":0.5265567302703857},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4836888909339905},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.472257524728775},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.448725163936615},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4036848545074463},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.36116915941238403},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.34182554483413696},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.17220449447631836},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.15436604619026184},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp39728.2021.9414572","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp39728.2021.9414572","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Peace, Justice and strong institutions","score":0.5400000214576721,"id":"https://metadata.un.org/sdg/16"}],"awards":[],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"},{"id":"https://openalex.org/F4320335787","display_name":"Fundamental Research Funds for the Central Universities","ror":null}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":55,"referenced_works":["https://openalex.org/W1724438581","https://openalex.org/W1821462560","https://openalex.org/W2113459411","https://openalex.org/W2119144962","https://openalex.org/W2170240176","https://openalex.org/W2896457183","https://openalex.org/W2923014074","https://openalex.org/W2924902521","https://openalex.org/W2944701285","https://openalex.org/W2950784811","https://openalex.org/W2951244744","https://openalex.org/W2962677625","https://openalex.org/W2963012544","https://openalex.org/W2963310665","https://openalex.org/W2963341956","https://openalex.org/W2964299589","https://openalex.org/W2965373594","https://openalex.org/W2970120757","https://openalex.org/W2970454332","https://openalex.org/W2970565456","https://openalex.org/W2970597249","https://openalex.org/W2975381464","https://openalex.org/W2978017171","https://openalex.org/W2994749257","https://openalex.org/W2996159613","https://openalex.org/W2996428491","https://openalex.org/W2997710335","https://openalex.org/W2998183051","https://openalex.org/W3012024854","https://openalex.org/W3014568172","https://openalex.org/W3034292689","https://openalex.org/W3035030897","https://openalex.org/W3035038672","https://openalex.org/W3038012435","https://openalex.org/W3105966348","https://openalex.org/W3177265267","https://openalex.org/W4288347855","https://openalex.org/W6637709462","https://openalex.org/W6638523607","https://openalex.org/W6676984168","https://openalex.org/W6677580257","https://openalex.org/W6685053522","https://openalex.org/W6750615492","https://openalex.org/W6755207826","https://openalex.org/W6760732026","https://openalex.org/W6762100871","https://openalex.org/W6762945437","https://openalex.org/W6763701032","https://openalex.org/W6766673545","https://openalex.org/W6767064347","https://openalex.org/W6768021236","https://openalex.org/W6768080748","https://openalex.org/W6768086466","https://openalex.org/W6768851824","https://openalex.org/W6775248542"],"related_works":["https://openalex.org/W2058965144","https://openalex.org/W2164382479","https://openalex.org/W2146343568","https://openalex.org/W98480971","https://openalex.org/W2150291671","https://openalex.org/W2013643406","https://openalex.org/W2027972911","https://openalex.org/W2097707447","https://openalex.org/W1966837078","https://openalex.org/W380079223"],"abstract_inverted_index":{"Despite":[0],"the":[1,26,52,74,79,88,143,150,153,160],"great":[2],"success":[3],"in":[4,165],"Natural":[5],"Language":[6],"Processing":[7],"(NLP)":[8],"area,":[9],"large":[10,27],"pre-trained":[11],"language":[12],"models":[13],"like":[14],"BERT":[15,39,141],"are":[16],"not":[17],"well-suited":[18],"for":[19,139],"resource-constrained":[20],"or":[21,99],"real-time":[22],"applications":[23],"owing":[24],"to":[25,84,87,116,123,148],"number":[28,53],"of":[29,54,152,163],"parameters":[30,55,98],"and":[31,37],"slow":[32],"inference":[33,67,81,111],"speed.":[34],"Recently,":[35],"compressing":[36],"accelerating":[38,140],"have":[40],"become":[41],"important":[42],"topics.":[43],"By":[44],"incorporating":[45],"a":[46,65],"parameter-sharing":[47],"strategy,":[48],"ALBERT":[49,61,85],"greatly":[50],"reduces":[51],"while":[56],"achieving":[57],"competitive":[58],"performance.":[59],"Nevertheless,":[60],"still":[62],"suffers":[63],"from":[64,114],"long":[66],"time.":[68],"In":[69],"this":[70],"work,":[71],"we":[72,157],"propose":[73],"ELBERT,":[75],"which":[76],"significantly":[77],"improves":[78],"average":[80],"speed":[82],"compared":[83,122],"due":[86],"proposed":[89],"confidence-window":[90],"based":[91],"early":[92,135,154],"exit":[93,136,155],"mechanism,":[94,156],"without":[95],"introducing":[96],"additional":[97],"extra":[100],"training":[101],"overhead.":[102],"Experimental":[103],"results":[104],"show":[105],"that":[106],"ELBERT":[107,129],"achieves":[108,130],"an":[109],"adaptive":[110],"speedup":[112],"varying":[113],"2\u00d7":[115],"10\u00d7":[117],"with":[118],"negligible":[119],"accuracy":[120,132],"degradation":[121],"AL-BERT":[124],"on":[125],"various":[126],"datasets.":[127],"Besides,":[128],"higher":[131],"than":[133],"existing":[134],"methods":[137],"used":[138],"under":[142],"same":[144],"computation":[145],"cost.":[146],"Furthermore,":[147],"understand":[149],"principle":[151],"also":[158],"visualize":[159],"decision-making":[161],"process":[162],"it":[164],"ELBERT.":[166],"Our":[167],"code":[168],"is":[169],"publicly":[170],"available":[171],"online.1":[172]},"counts_by_year":[{"year":2025,"cited_by_count":3},{"year":2024,"cited_by_count":5},{"year":2023,"cited_by_count":8},{"year":2022,"cited_by_count":1},{"year":2020,"cited_by_count":1}],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}