{"id":"https://openalex.org/W4200296936","doi":"https://doi.org/10.1109/iccad51958.2021.9643433","title":"Deferred Dropout: An Algorithm-Hardware Co-Design DNN Training Method Provisioning Consistent High Activation Sparsity","display_name":"Deferred Dropout: An Algorithm-Hardware Co-Design DNN Training Method Provisioning Consistent High Activation Sparsity","publication_year":2021,"publication_date":"2021-11-01","ids":{"openalex":"https://openalex.org/W4200296936","doi":"https://doi.org/10.1109/iccad51958.2021.9643433"},"language":"en","primary_location":{"id":"doi:10.1109/iccad51958.2021.9643433","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iccad51958.2021.9643433","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2021 IEEE/ACM International Conference On Computer Aided Design (ICCAD)","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/A5076633475","display_name":"Kangkyu Park","orcid":"https://orcid.org/0000-0003-1307-2735"},"institutions":[{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Kangkyu Park","raw_affiliation_strings":["KAIST,School of Electrical Engineering","School of Electrical Engineering, KAIST"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"KAIST,School of Electrical Engineering","institution_ids":["https://openalex.org/I157485424"]},{"raw_affiliation_string":"School of Electrical Engineering, KAIST","institution_ids":["https://openalex.org/I157485424"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5036617506","display_name":"Yunki Han","orcid":"https://orcid.org/0000-0003-0432-9324"},"institutions":[{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Yunki Han","raw_affiliation_strings":["KAIST,School of Electrical Engineering","School of Electrical Engineering, KAIST"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"KAIST,School of Electrical Engineering","institution_ids":["https://openalex.org/I157485424"]},{"raw_affiliation_string":"School of Electrical Engineering, KAIST","institution_ids":["https://openalex.org/I157485424"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5052390471","display_name":"Lee\u2010Sup Kim","orcid":"https://orcid.org/0000-0001-9585-4591"},"institutions":[{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Lee-Sup Kim","raw_affiliation_strings":["KAIST,School of Electrical Engineering","School of Electrical Engineering, KAIST"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"KAIST,School of Electrical Engineering","institution_ids":["https://openalex.org/I157485424"]},{"raw_affiliation_string":"School of Electrical Engineering, KAIST","institution_ids":["https://openalex.org/I157485424"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I157485424"],"apc_list":null,"apc_paid":null,"fwci":0.0955,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.40670712,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":90,"max":94},"biblio":{"volume":"32","issue":null,"first_page":"1","last_page":"9"},"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/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9994999766349792,"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"}},{"id":"https://openalex.org/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9983000159263611,"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/computer-science","display_name":"Computer science","score":0.8205514550209045},{"id":"https://openalex.org/keywords/memory-footprint","display_name":"Memory footprint","score":0.742292046546936},{"id":"https://openalex.org/keywords/dropout","display_name":"Dropout (neural networks)","score":0.7266668081283569},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.6167915463447571},{"id":"https://openalex.org/keywords/activation-function","display_name":"Activation function","score":0.57096266746521},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.5669408440589905},{"id":"https://openalex.org/keywords/reduction","display_name":"Reduction (mathematics)","score":0.5026659965515137},{"id":"https://openalex.org/keywords/provisioning","display_name":"Provisioning","score":0.44241416454315186},{"id":"https://openalex.org/keywords/performance-improvement","display_name":"Performance improvement","score":0.4396221339702606},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.3725016713142395},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.37169551849365234},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.3603745102882385},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.35554343461990356},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.31522440910339355},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.28820371627807617},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.17316067218780518},{"id":"https://openalex.org/keywords/computer-network","display_name":"Computer network","score":0.08807215094566345}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8205514550209045},{"id":"https://openalex.org/C74912251","wikidata":"https://www.wikidata.org/wiki/Q6815727","display_name":"Memory footprint","level":2,"score":0.742292046546936},{"id":"https://openalex.org/C2776145597","wikidata":"https://www.wikidata.org/wiki/Q25339462","display_name":"Dropout (neural networks)","level":2,"score":0.7266668081283569},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.6167915463447571},{"id":"https://openalex.org/C38365724","wikidata":"https://www.wikidata.org/wiki/Q4677469","display_name":"Activation function","level":3,"score":0.57096266746521},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.5669408440589905},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.5026659965515137},{"id":"https://openalex.org/C172191483","wikidata":"https://www.wikidata.org/wiki/Q1071806","display_name":"Provisioning","level":2,"score":0.44241416454315186},{"id":"https://openalex.org/C2778915421","wikidata":"https://www.wikidata.org/wiki/Q3643177","display_name":"Performance improvement","level":2,"score":0.4396221339702606},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.3725016713142395},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.37169551849365234},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.3603745102882385},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.35554343461990356},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.31522440910339355},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.28820371627807617},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.17316067218780518},{"id":"https://openalex.org/C31258907","wikidata":"https://www.wikidata.org/wiki/Q1301371","display_name":"Computer network","level":1,"score":0.08807215094566345},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.0},{"id":"https://openalex.org/C21547014","wikidata":"https://www.wikidata.org/wiki/Q1423657","display_name":"Operations management","level":1,"score":0.0},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","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/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/iccad51958.2021.9643433","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iccad51958.2021.9643433","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2021 IEEE/ACM International Conference On Computer Aided Design (ICCAD)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.5099999904632568,"display_name":"Quality Education","id":"https://metadata.un.org/sdg/4"}],"awards":[{"id":"https://openalex.org/G3757671995","display_name":null,"funder_award_id":"NRF-2020R1A2B5B02002690","funder_id":"https://openalex.org/F4320322120","funder_display_name":"National Research Foundation of Korea"}],"funders":[{"id":"https://openalex.org/F4320322120","display_name":"National Research Foundation of Korea","ror":"https://ror.org/013aysd81"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":54,"referenced_works":["https://openalex.org/W1921523184","https://openalex.org/W2095705004","https://openalex.org/W2168231600","https://openalex.org/W2338908902","https://openalex.org/W2489529491","https://openalex.org/W2516141709","https://openalex.org/W2606722458","https://openalex.org/W2622263826","https://openalex.org/W2625457103","https://openalex.org/W2701971652","https://openalex.org/W2750173518","https://openalex.org/W2773689216","https://openalex.org/W2883283076","https://openalex.org/W2896457183","https://openalex.org/W2899663614","https://openalex.org/W2900228909","https://openalex.org/W2913839334","https://openalex.org/W2936672709","https://openalex.org/W2947644971","https://openalex.org/W2949866178","https://openalex.org/W2962821792","https://openalex.org/W2963674932","https://openalex.org/W2964261616","https://openalex.org/W2980186997","https://openalex.org/W3034714646","https://openalex.org/W3036878841","https://openalex.org/W3037639655","https://openalex.org/W3043729931","https://openalex.org/W3043802286","https://openalex.org/W3092357178","https://openalex.org/W3092581319","https://openalex.org/W3100859887","https://openalex.org/W3129831491","https://openalex.org/W3157218387","https://openalex.org/W4287906222","https://openalex.org/W4288257146","https://openalex.org/W4289300273","https://openalex.org/W4292779060","https://openalex.org/W4295312788","https://openalex.org/W4385245566","https://openalex.org/W6638632666","https://openalex.org/W6674330103","https://openalex.org/W6684859321","https://openalex.org/W6739901393","https://openalex.org/W6740114376","https://openalex.org/W6754807246","https://openalex.org/W6755207826","https://openalex.org/W6756592394","https://openalex.org/W6759163087","https://openalex.org/W6766978945","https://openalex.org/W6768723914","https://openalex.org/W6772013979","https://openalex.org/W6778883912","https://openalex.org/W6779977557"],"related_works":["https://openalex.org/W3082178636","https://openalex.org/W3110577345","https://openalex.org/W2017432143","https://openalex.org/W2782041652","https://openalex.org/W1973694374","https://openalex.org/W2983574358","https://openalex.org/W2370475531","https://openalex.org/W2612657834","https://openalex.org/W4385924768","https://openalex.org/W2392157706"],"abstract_inverted_index":{"This":[0,162],"paper":[1],"proposes":[2],"a":[3,102,177],"deep":[4],"neural":[5],"network":[6],"training":[7,24,33,104,172,198],"method":[8,209],"that":[9,169,206],"provisions":[10],"consistent":[11],"high":[12,179],"activation":[13,37,90,201],"sparsity":[14,38,56,91],"and":[15,61,92,108,158,200,216],"the":[16,20,29,44,48,54,59,63,82,85,93,115,124,131,141,147,150,153,185,195,207,221,226],"ability":[17],"to":[18,43,130,193,225],"adjust":[19],"sparsity.":[21,202],"To":[22],"improve":[23],"performance,":[25],"prior":[26,227],"work":[27],"reduces":[28],"memory":[30,213],"footprint":[31,64,214],"for":[32,72],"by":[34,58],"exploiting":[35],"input":[36,118],"which":[39],"is":[40,66,144,164],"observed":[41],"due":[42],"ReLU":[45],"function.":[46],"However,":[47],"previous":[49,94],"approach":[50,95,163],"relies":[51],"solely":[52],"on":[53,166,220],"inherent":[55],"caused":[57],"function,":[60,83],"thus":[62],"reduction":[65,215],"not":[67,80],"guaranteed.":[68],"In":[69,99,128],"particular,":[70],"models":[71,86],"natural":[73],"language":[74],"processing":[75],"tasks":[76],"like":[77],"BERT":[78],"do":[79],"use":[81],"so":[84],"have":[87],"almost":[88],"zero":[89],"loses":[96],"its":[97,109],"efficiency.":[98],"this":[100],"paper,":[101],"new":[103],"method,":[105,117],"Deferred":[106],"Dropout,":[107],"hardware":[110,189],"architecture":[111,190],"are":[112,120,136,156,182],"proposed.":[113],"With":[114],"proposed":[116,208],"activations":[119,135,155,181],"dropped":[121],"out":[122],"after":[123],"conventional":[125,132],"forward-pass":[126,139],"computation.":[127,151],"contrast":[129],"dropout":[133],"where":[134],"zeroed":[137],"before":[138],"computation,":[140],"dropping":[142],"timing":[143],"deferred":[145],"until":[146],"completion":[148],"of":[149],"Then,":[152],"sparsified":[154],"compressed":[157],"stashed":[159],"in":[160,184],"memory.":[161],"based":[165],"our":[167],"observation":[168],"networks":[170],"preserve":[171],"quality":[173,199],"even":[174],"if":[175],"only":[176],"few":[178],"magnitude":[180],"used":[183],"backward":[186],"pass.":[187],"The":[188],"enables":[191],"designers":[192],"exploit":[194],"tradeoff":[196],"between":[197],"Evaluation":[203],"results":[204],"demonstrate":[205],"achieves":[210],"1.21-3.60":[211],"\u00d7":[212],"1.06-1.43":[217],"x":[218],"speedup":[219],"TPUv3":[222],"architecture,":[223],"compared":[224],"work.":[228]},"counts_by_year":[{"year":2023,"cited_by_count":1}],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}