{"id":"https://openalex.org/W4386699354","doi":"https://doi.org/10.1109/tmc.2023.3315138","title":"DeViT: Decomposing Vision Transformers for Collaborative Inference in Edge Devices","display_name":"DeViT: Decomposing Vision Transformers for Collaborative Inference in Edge Devices","publication_year":2023,"publication_date":"2023-09-13","ids":{"openalex":"https://openalex.org/W4386699354","doi":"https://doi.org/10.1109/tmc.2023.3315138"},"language":"en","primary_location":{"id":"doi:10.1109/tmc.2023.3315138","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tmc.2023.3315138","pdf_url":null,"source":{"id":"https://openalex.org/S69141925","display_name":"IEEE Transactions on Mobile Computing","issn_l":"1536-1233","issn":["1536-1233","1558-0660","2161-9875"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320439","host_organization_name":"IEEE Computer Society","host_organization_lineage":["https://openalex.org/P4310320439","https://openalex.org/P4310319808"],"host_organization_lineage_names":["IEEE Computer Society","Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Mobile Computing","raw_type":"journal-article"},"type":"article","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/A5047347817","display_name":"Guanyu Xu","orcid":"https://orcid.org/0000-0002-2771-9272"},"institutions":[{"id":"https://openalex.org/I125839683","display_name":"Beijing Institute of Technology","ror":"https://ror.org/01skt4w74","country_code":"CN","type":"education","lineage":["https://openalex.org/I125839683","https://openalex.org/I890469752"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Guanyu Xu","raw_affiliation_strings":["School of Information and Electrionics, Beijing Institute of Technology, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-2771-9272","affiliations":[{"raw_affiliation_string":"School of Information and Electrionics, Beijing Institute of Technology, Beijing, China","institution_ids":["https://openalex.org/I125839683"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5057198274","display_name":"Zhiwei Hao","orcid":"https://orcid.org/0000-0002-6237-7028"},"institutions":[{"id":"https://openalex.org/I125839683","display_name":"Beijing Institute of Technology","ror":"https://ror.org/01skt4w74","country_code":"CN","type":"education","lineage":["https://openalex.org/I125839683","https://openalex.org/I890469752"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhiwei Hao","raw_affiliation_strings":["School of Information and Electrionics, Beijing Institute of Technology, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-6237-7028","affiliations":[{"raw_affiliation_string":"School of Information and Electrionics, Beijing Institute of Technology, Beijing, China","institution_ids":["https://openalex.org/I125839683"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5039158745","display_name":"Yong Luo","orcid":"https://orcid.org/0000-0002-2296-6370"},"institutions":[{"id":"https://openalex.org/I37461747","display_name":"Wuhan University","ror":"https://ror.org/033vjfk17","country_code":"CN","type":"education","lineage":["https://openalex.org/I37461747"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yong Luo","raw_affiliation_strings":["School of Computer Science, Wuhan University, Wuhan, China"],"raw_orcid":"https://orcid.org/0000-0002-2296-6370","affiliations":[{"raw_affiliation_string":"School of Computer Science, Wuhan University, Wuhan, China","institution_ids":["https://openalex.org/I37461747"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5020194895","display_name":"Han Hu","orcid":"https://orcid.org/0000-0001-7532-0496"},"institutions":[{"id":"https://openalex.org/I125839683","display_name":"Beijing Institute of Technology","ror":"https://ror.org/01skt4w74","country_code":"CN","type":"education","lineage":["https://openalex.org/I125839683","https://openalex.org/I890469752"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Han Hu","raw_affiliation_strings":["School of Information and Electrionics, Beijing Institute of Technology, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0001-7532-0496","affiliations":[{"raw_affiliation_string":"School of Information and Electrionics, Beijing Institute of Technology, Beijing, China","institution_ids":["https://openalex.org/I125839683"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044674658","display_name":"Jianping An","orcid":"https://orcid.org/0000-0002-6441-9711"},"institutions":[{"id":"https://openalex.org/I125839683","display_name":"Beijing Institute of Technology","ror":"https://ror.org/01skt4w74","country_code":"CN","type":"education","lineage":["https://openalex.org/I125839683","https://openalex.org/I890469752"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jianping An","raw_affiliation_strings":["School of Information and Electrionics, Beijing Institute of Technology, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-6441-9711","affiliations":[{"raw_affiliation_string":"School of Information and Electrionics, Beijing Institute of Technology, Beijing, China","institution_ids":["https://openalex.org/I125839683"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5080122431","display_name":"Shiwen Mao","orcid":"https://orcid.org/0000-0002-7052-0007"},"institutions":[{"id":"https://openalex.org/I82497590","display_name":"Auburn University","ror":"https://ror.org/02v80fc35","country_code":"US","type":"education","lineage":["https://openalex.org/I82497590"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Shiwen Mao","raw_affiliation_strings":["Department of Electrical and Computer Engineering, Auburn University, Auburn, AL, USA"],"raw_orcid":"https://orcid.org/0000-0002-7052-0007","affiliations":[{"raw_affiliation_string":"Department of Electrical and Computer Engineering, Auburn University, Auburn, AL, USA","institution_ids":["https://openalex.org/I82497590"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":5.0125,"has_fulltext":false,"cited_by_count":51,"citation_normalized_percentile":{"value":0.96866957,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":99,"max":100},"biblio":{"volume":"23","issue":"5","first_page":"5917","last_page":"5932"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9998999834060669,"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":0.9998999834060669,"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":0.9969000220298767,"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/T11605","display_name":"Visual Attention and Saliency Detection","score":0.9934999942779541,"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/computer-science","display_name":"Computer science","score":0.8244612216949463},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.7628682255744934},{"id":"https://openalex.org/keywords/software-deployment","display_name":"Software deployment","score":0.5560773015022278},{"id":"https://openalex.org/keywords/edge-device","display_name":"Edge device","score":0.4674927294254303},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4459201693534851},{"id":"https://openalex.org/keywords/edge-computing","display_name":"Edge computing","score":0.4452781677246094},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.43699201941490173},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.4153619408607483},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.40128013491630554},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.37200087308883667},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.33386391401290894},{"id":"https://openalex.org/keywords/data-mining","display_name":"Data mining","score":0.32464635372161865},{"id":"https://openalex.org/keywords/enhanced-data-rates-for-gsm-evolution","display_name":"Enhanced Data Rates for GSM Evolution","score":0.3144398331642151},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.1974944770336151},{"id":"https://openalex.org/keywords/cloud-computing","display_name":"Cloud computing","score":0.14105895161628723},{"id":"https://openalex.org/keywords/software-engineering","display_name":"Software engineering","score":0.11474829912185669}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8244612216949463},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.7628682255744934},{"id":"https://openalex.org/C105339364","wikidata":"https://www.wikidata.org/wiki/Q2297740","display_name":"Software deployment","level":2,"score":0.5560773015022278},{"id":"https://openalex.org/C138236772","wikidata":"https://www.wikidata.org/wiki/Q25098575","display_name":"Edge device","level":3,"score":0.4674927294254303},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4459201693534851},{"id":"https://openalex.org/C2778456923","wikidata":"https://www.wikidata.org/wiki/Q5337692","display_name":"Edge computing","level":3,"score":0.4452781677246094},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.43699201941490173},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4153619408607483},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.40128013491630554},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.37200087308883667},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.33386391401290894},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.32464635372161865},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.3144398331642151},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.1974944770336151},{"id":"https://openalex.org/C79974875","wikidata":"https://www.wikidata.org/wiki/Q483639","display_name":"Cloud computing","level":2,"score":0.14105895161628723},{"id":"https://openalex.org/C115903868","wikidata":"https://www.wikidata.org/wiki/Q80993","display_name":"Software engineering","level":1,"score":0.11474829912185669},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/tmc.2023.3315138","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tmc.2023.3315138","pdf_url":null,"source":{"id":"https://openalex.org/S69141925","display_name":"IEEE Transactions on Mobile Computing","issn_l":"1536-1233","issn":["1536-1233","1558-0660","2161-9875"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320439","host_organization_name":"IEEE Computer Society","host_organization_lineage":["https://openalex.org/P4310320439","https://openalex.org/P4310319808"],"host_organization_lineage_names":["IEEE Computer Society","Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Transactions on Mobile Computing","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.6899999976158142,"display_name":"Affordable and clean energy","id":"https://metadata.un.org/sdg/7"}],"awards":[{"id":"https://openalex.org/G1284973545","display_name":null,"funder_award_id":"62276195","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G1865835052","display_name":"\u591a\u536b\u661f\u73af\u5883\u4e0b\u7684\u81ea\u5b66\u4e60\u6df1\u5ea6\u795e\u7ecf\u7f51\u7edc\u89e3\u8026\u7814\u7a76","funder_award_id":"61971457","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":51,"referenced_works":["https://openalex.org/W2108598243","https://openalex.org/W2138011018","https://openalex.org/W2533598788","https://openalex.org/W2605258629","https://openalex.org/W2896457183","https://openalex.org/W2931092525","https://openalex.org/W2946794439","https://openalex.org/W2949628230","https://openalex.org/W2952627098","https://openalex.org/W2962911728","https://openalex.org/W2963420686","https://openalex.org/W2972324944","https://openalex.org/W2973727699","https://openalex.org/W2991040477","https://openalex.org/W3018955265","https://openalex.org/W3034457371","https://openalex.org/W3035321581","https://openalex.org/W3046289513","https://openalex.org/W3047565185","https://openalex.org/W3094502228","https://openalex.org/W3100229168","https://openalex.org/W3105504538","https://openalex.org/W3105966348","https://openalex.org/W3109233295","https://openalex.org/W3118608800","https://openalex.org/W3121523901","https://openalex.org/W3138516171","https://openalex.org/W3159778524","https://openalex.org/W3172464367","https://openalex.org/W3173381523","https://openalex.org/W3174544005","https://openalex.org/W3194959296","https://openalex.org/W3196974791","https://openalex.org/W3204801262","https://openalex.org/W3204998121","https://openalex.org/W4281554877","https://openalex.org/W4281571140","https://openalex.org/W4285261368","https://openalex.org/W4385245566","https://openalex.org/W6742075157","https://openalex.org/W6762945437","https://openalex.org/W6767997687","https://openalex.org/W6775706467","https://openalex.org/W6787972765","https://openalex.org/W6788135285","https://openalex.org/W6794345597","https://openalex.org/W6795062860","https://openalex.org/W6800633732","https://openalex.org/W6802648153","https://openalex.org/W6838633097","https://openalex.org/W6840352845"],"related_works":["https://openalex.org/W4322761281","https://openalex.org/W4238233472","https://openalex.org/W3111395152","https://openalex.org/W4313526662","https://openalex.org/W4313463218","https://openalex.org/W3106131444","https://openalex.org/W3216099748","https://openalex.org/W4205963435","https://openalex.org/W4312996489","https://openalex.org/W3214037210"],"abstract_inverted_index":{"Recent":[0],"years":[1],"have":[2],"witnessed":[3],"the":[4,66,73,161,167,221,226,231,254],"great":[5],"success":[6],"of":[7,28,68,163],"vision":[8,19],"transformer":[9,69],"(ViT),":[10],"which":[11],"has":[12],"achieved":[13],"state-of-the-art":[14],"performance":[15],"on":[16,38,133,177,225,240,253],"multiple":[17,77,140],"computer":[18],"benchmarks.":[20],"However,":[21],"ViT":[22,46,75,175],"models":[23,47,79,143,152,165],"suffer":[24],"from":[25,166],"vast":[26],"amounts":[27],"parameters":[29],"and":[30,71,92,149,190,201,247],"high":[31],"computation":[32],"cost,":[33],"leading":[34],"to":[35,48,64,89,118,138,159,220],"difficult":[36],"deployment":[37,121],"resource-constrained":[39],"edge":[40,84,120,255],"devices.":[41,85],"Existing":[42],"solutions":[43],"mostly":[44],"compress":[45],"a":[49,110,129,155,196],"compact":[50],"model":[51],"but":[52],"still":[53],"cannot":[54],"achieve":[55,90],"real-time":[56],"inference.":[57],"To":[58,104],"tackle":[59],"this":[60,105],"issue,":[61],"we":[62,107,127],"propose":[63,109],"explore":[65],"divisibility":[67],"structure,":[70],"decompose":[72],"large":[74,102,124,168,222],"into":[76],"small":[78,141],"for":[80,172,188],"collaborative":[81,94,111,186],"inference":[82,95,112,187],"at":[83],"Our":[86],"objective":[87],"is":[88],"fast":[91],"energy-efficient":[93],"while":[96,144,243],"maintaining":[97],"comparable":[98],"accuracy":[99,215,239],"compared":[100,219],"with":[101,212],"ViTs.":[103,125],"end,":[106],"first":[108],"framework":[113],"termed":[114,136],"<bold":[115],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[116],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">DeViT</b>":[117],"facilitate":[119],"by":[122,153,210,236],"decomposing":[123],"Subsequently,":[126],"design":[128],"decomposition-and-ensemble":[130],"algorithm":[131],"based":[132],"knowledge":[134],"distillation,":[135],"DEKD,":[137],"fuse":[139],"decomposed":[142,164],"dramatically":[145],"reducing":[146],"communication":[147],"overheads,":[148],"handle":[150],"heterogeneous":[151],"developing":[154],"feature":[156],"matching":[157],"module":[158],"promote":[160],"imitations":[162],"ViT.":[169],"Extensive":[170],"experiments":[171],"three":[173],"representative":[174],"backbones":[176],"four":[178],"widely-used":[179],"datasets":[180],"demonstrate":[181],"our":[182,205],"method":[183],"achieves":[184],"efficient":[185,233],"ViTs":[189],"outperforms":[191],"existing":[192],"lightweight":[193],"ViTs,":[194],"striking":[195],"good":[197],"trade-off":[198],"between":[199],"efficiency":[200],"accuracy.":[202],"For":[203],"example,":[204],"DeViTs":[206],"improves":[207],"end-to-end":[208],"latency":[209],"2.89\u00d7":[211],"only":[213],"1.65%":[214],"sacrifice":[216],"using":[217],"CIFAR-100":[218],"ViT,":[223,234],"ViT-L/16,":[224],"GPU":[227],"server.":[228],"DeDeiTs":[229],"surpasses":[230],"recent":[232],"MobileViT-S,":[235],"3.54%":[237],"in":[238],"ImageNet-1":[241],"K,":[242],"running":[244],"1.72\u00d7":[245],"faster":[246],"requiring":[248],"55.28%":[249],"lower":[250],"energy":[251],"consumption":[252],"device.":[256]},"counts_by_year":[{"year":2026,"cited_by_count":10},{"year":2025,"cited_by_count":32},{"year":2024,"cited_by_count":9}],"updated_date":"2026-07-15T18:14:33.161393","created_date":"2025-10-10T00:00:00"}