{"id":"https://openalex.org/W3199463158","doi":"https://doi.org/10.1145/3477008","title":"Exploiting Activation Sparsity for Fast CNN Inference on Mobile GPUs","display_name":"Exploiting Activation Sparsity for Fast CNN Inference on Mobile GPUs","publication_year":2021,"publication_date":"2021-09-17","ids":{"openalex":"https://openalex.org/W3199463158","doi":"https://doi.org/10.1145/3477008","mag":"3199463158"},"language":"en","primary_location":{"id":"doi:10.1145/3477008","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3477008","pdf_url":null,"source":{"id":"https://openalex.org/S136160450","display_name":"ACM Transactions on Embedded Computing Systems","issn_l":"1539-9087","issn":["1539-9087","1558-3465"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM Transactions on Embedded Computing Systems","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/A5063943906","display_name":"Chanyoung Oh","orcid":"https://orcid.org/0000-0003-2877-1141"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Chanyoung Oh","raw_affiliation_strings":["KT AI2XL, Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"KT AI2XL, Republic of Korea","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5032614836","display_name":"Junhyuk So","orcid":"https://orcid.org/0000-0002-9210-1284"},"institutions":[{"id":"https://openalex.org/I124633538","display_name":"University of Seoul","ror":"https://ror.org/05en5nh73","country_code":"KR","type":"education","lineage":["https://openalex.org/I124633538"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Junhyuk So","raw_affiliation_strings":["University of Seoul, Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Seoul, Republic of Korea","institution_ids":["https://openalex.org/I124633538"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100429098","display_name":"Sumin Kim","orcid":"https://orcid.org/0000-0003-2278-5278"},"institutions":[{"id":"https://openalex.org/I124633538","display_name":"University of Seoul","ror":"https://ror.org/05en5nh73","country_code":"KR","type":"education","lineage":["https://openalex.org/I124633538"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Sumin Kim","raw_affiliation_strings":["University of Seoul, Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Seoul, Republic of Korea","institution_ids":["https://openalex.org/I124633538"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5029634385","display_name":"Youngmin Yi","orcid":"https://orcid.org/0000-0001-9802-2109"},"institutions":[{"id":"https://openalex.org/I124633538","display_name":"University of Seoul","ror":"https://ror.org/05en5nh73","country_code":"KR","type":"education","lineage":["https://openalex.org/I124633538"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Youngmin Yi","raw_affiliation_strings":["University of Seoul, Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"University of Seoul, Republic of Korea","institution_ids":["https://openalex.org/I124633538"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5063943906"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.4849,"has_fulltext":false,"cited_by_count":6,"citation_normalized_percentile":{"value":0.65406714,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":98},"biblio":{"volume":"20","issue":"5s","first_page":"1","last_page":"25"},"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":0.9948999881744385,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9915000200271606,"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":0.8892539143562317},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6635967493057251},{"id":"https://openalex.org/keywords/pruning","display_name":"Pruning","score":0.6459137797355652},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.6225989460945129},{"id":"https://openalex.org/keywords/convolution","display_name":"Convolution (computer science)","score":0.6163250803947449},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.5888330340385437},{"id":"https://openalex.org/keywords/mobile-device","display_name":"Mobile device","score":0.5630229711532593},{"id":"https://openalex.org/keywords/focus","display_name":"Focus (optics)","score":0.5330848693847656},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.5127569437026978},{"id":"https://openalex.org/keywords/kernel","display_name":"Kernel (algebra)","score":0.4710255265235901},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.4616905450820923},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.38234007358551025},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3737630844116211},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.3383112847805023},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.07213374972343445}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8892539143562317},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6635967493057251},{"id":"https://openalex.org/C108010975","wikidata":"https://www.wikidata.org/wiki/Q500094","display_name":"Pruning","level":2,"score":0.6459137797355652},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.6225989460945129},{"id":"https://openalex.org/C45347329","wikidata":"https://www.wikidata.org/wiki/Q5166604","display_name":"Convolution (computer science)","level":3,"score":0.6163250803947449},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.5888330340385437},{"id":"https://openalex.org/C186967261","wikidata":"https://www.wikidata.org/wiki/Q5082128","display_name":"Mobile device","level":2,"score":0.5630229711532593},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.5330848693847656},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.5127569437026978},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.4710255265235901},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.4616905450820923},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.38234007358551025},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3737630844116211},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3383112847805023},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.07213374972343445},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.0},{"id":"https://openalex.org/C114614502","wikidata":"https://www.wikidata.org/wiki/Q76592","display_name":"Combinatorics","level":1,"score":0.0},{"id":"https://openalex.org/C120665830","wikidata":"https://www.wikidata.org/wiki/Q14620","display_name":"Optics","level":1,"score":0.0},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"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/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.0},{"id":"https://openalex.org/C6557445","wikidata":"https://www.wikidata.org/wiki/Q173113","display_name":"Agronomy","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3477008","is_oa":false,"landing_page_url":"https://doi.org/10.1145/3477008","pdf_url":null,"source":{"id":"https://openalex.org/S136160450","display_name":"ACM Transactions on Embedded Computing Systems","issn_l":"1539-9087","issn":["1539-9087","1558-3465"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319798","host_organization_name":"Association for Computing Machinery","host_organization_lineage":["https://openalex.org/P4310319798"],"host_organization_lineage_names":["Association for Computing Machinery"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ACM Transactions on Embedded Computing Systems","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Industry, innovation and infrastructure","score":0.4699999988079071,"id":"https://metadata.un.org/sdg/9"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":25,"referenced_works":["https://openalex.org/W2097117768","https://openalex.org/W2108598243","https://openalex.org/W2126105956","https://openalex.org/W2172654076","https://openalex.org/W2194775991","https://openalex.org/W2347043188","https://openalex.org/W2516141709","https://openalex.org/W2562731582","https://openalex.org/W2750173518","https://openalex.org/W2772161496","https://openalex.org/W2924515500","https://openalex.org/W2948634622","https://openalex.org/W2949676527","https://openalex.org/W2962816593","https://openalex.org/W2963896595","https://openalex.org/W2968480143","https://openalex.org/W2971115712","https://openalex.org/W2982083293","https://openalex.org/W2996874060","https://openalex.org/W3006363700","https://openalex.org/W3034733718","https://openalex.org/W3035678286","https://openalex.org/W3090912412","https://openalex.org/W3104849992","https://openalex.org/W4244330903"],"related_works":["https://openalex.org/W2058965144","https://openalex.org/W2164382479","https://openalex.org/W98480971","https://openalex.org/W2150291671","https://openalex.org/W2027972911","https://openalex.org/W2146343568","https://openalex.org/W2013643406","https://openalex.org/W2157978810","https://openalex.org/W2597809628","https://openalex.org/W3157543420"],"abstract_inverted_index":{"Over":[0],"the":[1,5,32,38,51,62,102,114,152,172],"past":[2],"several":[3,119],"years,":[4],"need":[6],"for":[7,57],"on-device":[8],"deep":[9],"learning":[10],"has":[11,22,28],"been":[12,23,29],"rapidly":[13],"increasing,":[14],"and":[15,100,143,164],"efficient":[16,158],"CNN":[17,91],"inference":[18,92],"on":[19,42,50,64,93,151,184],"mobile":[20,94],"platforms":[21],"actively":[24],"researched.":[25],"Sparsity":[26],"exploitation":[27],"one":[30],"of":[31],"most":[33],"active":[34],"research":[35,63],"themes,":[36],"but":[37],"studies":[39],"mostly":[40],"focus":[41],"weight":[43,46],"sparsity":[44,66,99,127,142],"by":[45],"pruning.":[47],"Activation":[48],"sparsity,":[49],"contrary,":[52],"requires":[53],"compression":[54,115,163],"at":[55],"runtime":[56],"every":[58],"input":[59],"tensor.":[60],"Hence,":[61],"activation":[65,98,126],"mainly":[67],"targets":[68],"NPUs":[69],"that":[70,85,101,123,166],"can":[71,124,168],"efficiently":[72],"process":[73],"this":[74,81],"with":[75,96,138],"their":[76],"own":[77],"hardware":[78],"logic.":[79],"In":[80,131],"paper,":[82],"we":[83,133,155,177],"observe":[84],"it":[86,167],"is":[87,108],"difficult":[88],"to":[89,113,182],"accelerate":[90],"GPUs":[95],"natural":[97],"widely":[103],"used":[104],"CSR-based":[105],"sparse":[106,145,159],"convolution":[107,146,149,160],"not":[109],"sufficiently":[110],"effective":[111],"due":[112],"overhead.":[116],"We":[117],"propose":[118],"novel":[120],"sparsification":[121],"methods":[122],"boost":[125],"without":[128,162],"harming":[129],"accuracy.":[130],"particular,":[132],"selectively":[134],"sparsify":[135],"some":[136],"layers":[137],"an":[139,157],"extremely":[140],"high":[141],"adopt":[144],"or":[147],"dense":[148],"depending":[150],"layers.":[153],"Further,":[154],"present":[156],"method":[161],"demonstrate":[165],"be":[169],"faster":[170],"than":[171],"CSR":[173],"implementation.":[174],"With":[175],"ResNet-50,":[176],"achieved":[178],"1.88":[179],"speedup":[180],"compared":[181],"TFLite":[183],"a":[185],"Mali-G76":[186],"GPU.":[187]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2024,"cited_by_count":1},{"year":2023,"cited_by_count":2},{"year":2022,"cited_by_count":2}],"updated_date":"2026-05-21T06:26:12.895304","created_date":"2025-10-10T00:00:00"}