{"id":"https://openalex.org/W7138395053","doi":"https://doi.org/10.1609/aaai.v40i25.39225","title":"EnViT: Enhancing the Performance of Early-Exit Vision Transformers via Exit-Aware Structured Dropout-Enabled Self-Distillation","display_name":"EnViT: Enhancing the Performance of Early-Exit Vision Transformers via Exit-Aware Structured Dropout-Enabled Self-Distillation","publication_year":2026,"publication_date":"2026-03-14","ids":{"openalex":"https://openalex.org/W7138395053","doi":"https://doi.org/10.1609/aaai.v40i25.39225"},"language":null,"primary_location":{"id":"doi:10.1609/aaai.v40i25.39225","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i25.39225","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/39225/43186","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"diamond","oa_url":"https://ojs.aaai.org/index.php/AAAI/article/download/39225/43186","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5129658948","display_name":"Yonghao Dong","orcid":null},"institutions":[{"id":"https://openalex.org/I196699116","display_name":"Wuhan University of Technology","ror":"https://ror.org/03fe7t173","country_code":"CN","type":"education","lineage":["https://openalex.org/I196699116"]},{"id":"https://openalex.org/I47720641","display_name":"Huazhong University of Science and Technology","ror":"https://ror.org/00p991c53","country_code":"CN","type":"education","lineage":["https://openalex.org/I47720641"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Yonghao Dong","raw_affiliation_strings":["National Engineering Research Center for Big Data Technology and System, Wuhan, China\nServices Computing Technology and System Lab, Wuhan, China\nCluster and Grid Computing Lab, Wuhan, China\nSchool of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, China"],"affiliations":[{"raw_affiliation_string":"National Engineering Research Center for Big Data Technology and System, Wuhan, China\nServices Computing Technology and System Lab, Wuhan, China\nCluster and Grid Computing Lab, Wuhan, China\nSchool of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, China","institution_ids":["https://openalex.org/I47720641","https://openalex.org/I196699116"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129650975","display_name":"Qiang He","orcid":null},"institutions":[{"id":"https://openalex.org/I196699116","display_name":"Wuhan University of Technology","ror":"https://ror.org/03fe7t173","country_code":"CN","type":"education","lineage":["https://openalex.org/I196699116"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Qiang He","raw_affiliation_strings":["National Engineering Research Center for Big Data Technology and System, Wuhan, China\nServices Computing Technology and System Lab, Wuhan, China\nCluster and Grid Computing Lab, Wuhan, China\nSchool of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, China\nSwinburne University of Technology, Melbourne, Australia"],"affiliations":[{"raw_affiliation_string":"National Engineering Research Center for Big Data Technology and System, Wuhan, China\nServices Computing Technology and System Lab, Wuhan, China\nCluster and Grid Computing Lab, Wuhan, China\nSchool of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, China\nSwinburne University of Technology, Melbourne, Australia","institution_ids":["https://openalex.org/I196699116"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129731433","display_name":"Penghong Rui","orcid":null},"institutions":[{"id":"https://openalex.org/I196699116","display_name":"Wuhan University of Technology","ror":"https://ror.org/03fe7t173","country_code":"CN","type":"education","lineage":["https://openalex.org/I196699116"]},{"id":"https://openalex.org/I47720641","display_name":"Huazhong University of Science and Technology","ror":"https://ror.org/00p991c53","country_code":"CN","type":"education","lineage":["https://openalex.org/I47720641"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Penghong Rui","raw_affiliation_strings":["National Engineering Research Center for Big Data Technology and System, Wuhan, China\nServices Computing Technology and System Lab, Wuhan, China\nCluster and Grid Computing Lab, Wuhan, China\nSchool of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, China"],"affiliations":[{"raw_affiliation_string":"National Engineering Research Center for Big Data Technology and System, Wuhan, China\nServices Computing Technology and System Lab, Wuhan, China\nCluster and Grid Computing Lab, Wuhan, China\nSchool of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, China","institution_ids":["https://openalex.org/I47720641","https://openalex.org/I196699116"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129751099","display_name":"Zhenzhe Zheng","orcid":null},"institutions":[{"id":"https://openalex.org/I183067930","display_name":"Shanghai Jiao Tong University","ror":"https://ror.org/0220qvk04","country_code":"CN","type":"education","lineage":["https://openalex.org/I183067930"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhenzhe Zheng","raw_affiliation_strings":["Shanghai Jiao Tong University, Shanghai, China"],"affiliations":[{"raw_affiliation_string":"Shanghai Jiao Tong University, Shanghai, China","institution_ids":["https://openalex.org/I183067930"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129645995","display_name":"Zhao Li","orcid":null},"institutions":[{"id":"https://openalex.org/I4210123185","display_name":"Zhejiang Lab","ror":"https://ror.org/02m2h7991","country_code":"CN","type":"facility","lineage":["https://openalex.org/I4210123185"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhao Li","raw_affiliation_strings":["Zhejiang Lab, Hangzhou, China"],"affiliations":[{"raw_affiliation_string":"Zhejiang Lab, Hangzhou, China","institution_ids":["https://openalex.org/I4210123185"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5122882789","display_name":"Feifei Chen","orcid":null},"institutions":[{"id":"https://openalex.org/I149704539","display_name":"Deakin University","ror":"https://ror.org/02czsnj07","country_code":"AU","type":"education","lineage":["https://openalex.org/I149704539"]}],"countries":["AU"],"is_corresponding":false,"raw_author_name":"Feifei Chen","raw_affiliation_strings":["Deakin University, Melbourne, Australia"],"affiliations":[{"raw_affiliation_string":"Deakin University, Melbourne, Australia","institution_ids":["https://openalex.org/I149704539"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5129746825","display_name":"Hai Jin","orcid":null},"institutions":[{"id":"https://openalex.org/I196699116","display_name":"Wuhan University of Technology","ror":"https://ror.org/03fe7t173","country_code":"CN","type":"education","lineage":["https://openalex.org/I196699116"]},{"id":"https://openalex.org/I47720641","display_name":"Huazhong University of Science and Technology","ror":"https://ror.org/00p991c53","country_code":"CN","type":"education","lineage":["https://openalex.org/I47720641"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Hai Jin","raw_affiliation_strings":["National Engineering Research Center for Big Data Technology and System, Wuhan, China\nServices Computing Technology and System Lab, Wuhan, China\nCluster and Grid Computing Lab, Wuhan, China\nSchool of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, China"],"affiliations":[{"raw_affiliation_string":"National Engineering Research Center for Big Data Technology and System, Wuhan, China\nServices Computing Technology and System Lab, Wuhan, China\nCluster and Grid Computing Lab, Wuhan, China\nSchool of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, China","institution_ids":["https://openalex.org/I47720641","https://openalex.org/I196699116"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5129734406","display_name":"Yun Yang","orcid":null},"institutions":[{"id":"https://openalex.org/I57093077","display_name":"Swinburne University of Technology","ror":"https://ror.org/031rekg67","country_code":"AU","type":"education","lineage":["https://openalex.org/I57093077"]}],"countries":["AU"],"is_corresponding":false,"raw_author_name":"Yun Yang","raw_affiliation_strings":["Swinburne University of Technology, Melbourne, Australia"],"affiliations":[{"raw_affiliation_string":"Swinburne University of Technology, Melbourne, Australia","institution_ids":["https://openalex.org/I57093077"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":8,"corresponding_author_ids":["https://openalex.org/A5129658948"],"corresponding_institution_ids":["https://openalex.org/I196699116","https://openalex.org/I47720641"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.77089552,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"40","issue":"25","first_page":"20852","last_page":"20860"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.6614999771118164,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.6614999771118164,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.049800001084804535,"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/T10502","display_name":"Advanced Memory and Neural Computing","score":0.04969999939203262,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6877999901771545},{"id":"https://openalex.org/keywords/edge-device","display_name":"Edge device","score":0.474700003862381},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.4632999897003174},{"id":"https://openalex.org/keywords/ranging","display_name":"Ranging","score":0.4108999967575073},{"id":"https://openalex.org/keywords/offset","display_name":"Offset (computer science)","score":0.3707999885082245},{"id":"https://openalex.org/keywords/edge-computing","display_name":"Edge computing","score":0.36959999799728394}],"concepts":[{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6877999901771545},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6625999808311462},{"id":"https://openalex.org/C138236772","wikidata":"https://www.wikidata.org/wiki/Q25098575","display_name":"Edge device","level":3,"score":0.474700003862381},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.4632999897003174},{"id":"https://openalex.org/C115051666","wikidata":"https://www.wikidata.org/wiki/Q6522493","display_name":"Ranging","level":2,"score":0.4108999967575073},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.37619999051094055},{"id":"https://openalex.org/C175291020","wikidata":"https://www.wikidata.org/wiki/Q1156822","display_name":"Offset (computer science)","level":2,"score":0.3707999885082245},{"id":"https://openalex.org/C2778456923","wikidata":"https://www.wikidata.org/wiki/Q5337692","display_name":"Edge computing","level":3,"score":0.36959999799728394},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.36419999599456787},{"id":"https://openalex.org/C2776145597","wikidata":"https://www.wikidata.org/wiki/Q25339462","display_name":"Dropout (neural networks)","level":2,"score":0.35530000925064087},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.33730000257492065},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.33309999108314514},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.33219999074935913},{"id":"https://openalex.org/C200601418","wikidata":"https://www.wikidata.org/wiki/Q2193887","display_name":"Reliability engineering","level":1,"score":0.33160001039505005},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.33090001344680786},{"id":"https://openalex.org/C44154836","wikidata":"https://www.wikidata.org/wiki/Q45045","display_name":"Simulation","level":1,"score":0.30169999599456787},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.29269999265670776},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.2651999890804291},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.2587999999523163},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.25540000200271606}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1609/aaai.v40i25.39225","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i25.39225","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/39225/43186","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1609/aaai.v40i25.39225","is_oa":true,"landing_page_url":"https://doi.org/10.1609/aaai.v40i25.39225","pdf_url":"https://ojs.aaai.org/index.php/AAAI/article/download/39225/43186","source":{"id":"https://openalex.org/S4210191458","display_name":"Proceedings of the AAAI Conference on Artificial Intelligence","issn_l":"2159-5399","issn":["2159-5399","2374-3468"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/P4310320058","host_organization_name":"Association for the Advancement of Artificial Intelligence","host_organization_lineage":["https://openalex.org/P4310320058"],"host_organization_lineage_names":["Association for the Advancement of Artificial Intelligence"],"type":"conference"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the AAAI Conference on Artificial Intelligence","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7138395053.pdf","grobid_xml":"https://content.openalex.org/works/W7138395053.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Vision":[0],"Transformers":[1],"(ViTs)":[2],"have":[3],"gained":[4],"significant":[5],"attention":[6],"and":[7,33,46,70,142,165],"widespread":[8],"adoption":[9],"due":[10],"to":[11,29,103,131,162,194,203],"their":[12,23],"impressive":[13],"performance":[14,119,167],"in":[15,21,94,169],"various":[16],"computer":[17],"vision":[18],"tasks.":[19],"However,":[20],"practice,":[22],"substantial":[24,91],"computational":[25,72],"overhead":[26,73],"often":[27],"leads":[28],"high":[30],"inference":[31,63,69],"latency":[32],"increased":[34],"overheads":[35],"when":[36],"deployed":[37],"on":[38,64,82,181],"resource-constrained":[39],"edge":[40,65],"devices":[41],"like":[42],"smartphones,":[43],"autonomous":[44],"vehicles,":[45],"robots.":[47],"To":[48],"address":[49],"these":[50],"challenges,":[51],"Early":[52],"Exit":[53],"(EE)":[54],"has":[55],"emerged":[56],"as":[57,139,151],"a":[58],"promising":[59],"approach":[60,115],"for":[61],"lightweight":[62],"devices.":[66],"It":[67],"accelerates":[68],"reduces":[71],"by":[74,148,172],"adaptively":[75],"producing":[76],"predictions":[77],"through":[78],"early":[79,104,121,163],"exits":[80,96,122,164,171,177],"based":[81],"sample":[83],"complexity.":[84],"Existing":[85],"EE":[86],"methods":[87],"typically":[88],"suffer":[89],"from":[90,158,192],"accuracy":[92,101,189],"decreases":[93],"late":[95,125,170],"while":[97,196],"providing":[98],"only":[99],"marginal":[100],"improvements":[102,190],"exits.":[105,126],"This":[106,153],"paper":[107],"presents":[108],"EnViT,":[109],"an":[110],"exit-aware":[111],"structured":[112,129,149],"dropout-enabled":[113],"self-distillation":[114],"that":[116,185],"enhances":[117],"the":[118,135,140,159],"of":[120,201],"without":[123],"compromising":[124],"EnViT":[127,187],"leverages":[128],"dropout":[130,150],"enable":[132],"self-distillation,":[133],"where":[134],"full":[136,160],"model":[137,161],"serves":[138],"teacher":[141],"its":[143],"own":[144],"virtual":[145],"sub-models":[146],"generated":[147],"students.":[152],"mechanism":[154],"effectively":[155],"distills":[156],"knowledge":[157],"avoids":[166],"degradation":[168],"mitigating":[173],"parameter":[174],"conflicts":[175],"across":[176],"during":[178],"training.":[179],"Evaluation":[180],"five":[182],"datasets":[183],"shows":[184],"our":[186],"achieves":[188],"ranging":[191],"0.36%":[193],"7.92%":[195],"maintaining":[197],"competitive":[198],"speed-up":[199],"ratios":[200],"1.72x":[202],"2.23x.":[204]},"counts_by_year":[],"updated_date":"2026-03-20T20:47:17.329874","created_date":"2026-03-18T00:00:00"}