{"id":"https://openalex.org/W3173567919","doi":"https://doi.org/10.1109/emc2-nips53020.2019.00009","title":"Discovering Low-Precision Networks Close to Full-Precision Networks for Efficient Inference","display_name":"Discovering Low-Precision Networks Close to Full-Precision Networks for Efficient Inference","publication_year":2019,"publication_date":"2019-12-01","ids":{"openalex":"https://openalex.org/W3173567919","doi":"https://doi.org/10.1109/emc2-nips53020.2019.00009","mag":"3173567919"},"language":"en","primary_location":{"id":"doi:10.1109/emc2-nips53020.2019.00009","is_oa":false,"landing_page_url":"https://doi.org/10.1109/emc2-nips53020.2019.00009","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2019 Fifth Workshop on Energy Efficient Machine Learning and Cognitive Computing - NeurIPS Edition (EMC2-NIPS)","raw_type":"proceedings-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/A5059206462","display_name":"Jeffrey L. McKinstry","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jeffrey L. McKinstry","raw_affiliation_strings":["IBM Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5055044754","display_name":"Steven K. Esser","orcid":"https://orcid.org/0009-0008-4089-0577"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Steven K. Esser","raw_affiliation_strings":["IBM Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113665253","display_name":"Rathinakumar Appuswamy","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rathinakumar Appuswamy","raw_affiliation_strings":["IBM Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5045970662","display_name":"Deepika Bablani","orcid":"https://orcid.org/0000-0002-8987-8201"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Deepika Bablani","raw_affiliation_strings":["IBM Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5087078947","display_name":"John V. Arthur","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"John V. Arthur","raw_affiliation_strings":["IBM Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5087595691","display_name":"Izzet B. Yildiz","orcid":"https://orcid.org/0000-0002-7998-4449"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Izzet B. Yildiz","raw_affiliation_strings":["IBM Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5063718511","display_name":"Dharmendra S. Modha","orcid":"https://orcid.org/0009-0005-6714-8127"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Dharmendra S. Modha","raw_affiliation_strings":["IBM Research"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":7,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":2.1355,"has_fulltext":false,"cited_by_count":40,"citation_normalized_percentile":{"value":0.90492387,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":95,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"6","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":0.9987999796867371,"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":0.9987999796867371,"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":0.9987999796867371,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.9976999759674072,"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":0.7488692998886108},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.7055697441101074},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.690773606300354},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.652028501033783},{"id":"https://openalex.org/keywords/quadratic-growth","display_name":"Quadratic growth","score":0.5980361700057983},{"id":"https://openalex.org/keywords/baseline","display_name":"Baseline (sea)","score":0.5560829639434814},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.5103265643119812},{"id":"https://openalex.org/keywords/range","display_name":"Range (aeronautics)","score":0.5018899440765381},{"id":"https://openalex.org/keywords/variance","display_name":"Variance (accounting)","score":0.48549798130989075},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4591050148010254},{"id":"https://openalex.org/keywords/calibration","display_name":"Calibration","score":0.44115686416625977},{"id":"https://openalex.org/keywords/accuracy-and-precision","display_name":"Accuracy and precision","score":0.4397865831851959},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.4321553707122803},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.3938867151737213},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.15879207849502563},{"id":"https://openalex.org/keywords/statistics","display_name":"Statistics","score":0.14055508375167847}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7488692998886108},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.7055697441101074},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.690773606300354},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.652028501033783},{"id":"https://openalex.org/C195956108","wikidata":"https://www.wikidata.org/wiki/Q7268362","display_name":"Quadratic growth","level":2,"score":0.5980361700057983},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.5560829639434814},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.5103265643119812},{"id":"https://openalex.org/C204323151","wikidata":"https://www.wikidata.org/wiki/Q905424","display_name":"Range (aeronautics)","level":2,"score":0.5018899440765381},{"id":"https://openalex.org/C196083921","wikidata":"https://www.wikidata.org/wiki/Q7915758","display_name":"Variance (accounting)","level":2,"score":0.48549798130989075},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4591050148010254},{"id":"https://openalex.org/C165838908","wikidata":"https://www.wikidata.org/wiki/Q736777","display_name":"Calibration","level":2,"score":0.44115686416625977},{"id":"https://openalex.org/C202799725","wikidata":"https://www.wikidata.org/wiki/Q272035","display_name":"Accuracy and precision","level":2,"score":0.4397865831851959},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4321553707122803},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3938867151737213},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.15879207849502563},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.14055508375167847},{"id":"https://openalex.org/C111368507","wikidata":"https://www.wikidata.org/wiki/Q43518","display_name":"Oceanography","level":1,"score":0.0},{"id":"https://openalex.org/C13280743","wikidata":"https://www.wikidata.org/wiki/Q131089","display_name":"Geodesy","level":1,"score":0.0},{"id":"https://openalex.org/C159985019","wikidata":"https://www.wikidata.org/wiki/Q181790","display_name":"Composite material","level":1,"score":0.0},{"id":"https://openalex.org/C121955636","wikidata":"https://www.wikidata.org/wiki/Q4116214","display_name":"Accounting","level":1,"score":0.0},{"id":"https://openalex.org/C144133560","wikidata":"https://www.wikidata.org/wiki/Q4830453","display_name":"Business","level":0,"score":0.0},{"id":"https://openalex.org/C192562407","wikidata":"https://www.wikidata.org/wiki/Q228736","display_name":"Materials science","level":0,"score":0.0},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"score":0.0},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.0},{"id":"https://openalex.org/C127313418","wikidata":"https://www.wikidata.org/wiki/Q1069","display_name":"Geology","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/emc2-nips53020.2019.00009","is_oa":false,"landing_page_url":"https://doi.org/10.1109/emc2-nips53020.2019.00009","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2019 Fifth Workshop on Energy Efficient Machine Learning and Cognitive Computing - NeurIPS Edition (EMC2-NIPS)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/7","display_name":"Affordable and clean energy","score":0.9100000262260437}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":23,"referenced_works":["https://openalex.org/W1902934009","https://openalex.org/W2242818861","https://openalex.org/W2291160084","https://openalex.org/W2314470091","https://openalex.org/W2557283755","https://openalex.org/W2751477244","https://openalex.org/W2766164908","https://openalex.org/W2786771851","https://openalex.org/W2789142637","https://openalex.org/W2798693909","https://openalex.org/W2885688430","https://openalex.org/W2963122961","https://openalex.org/W2963521187","https://openalex.org/W2964203871","https://openalex.org/W2964228333","https://openalex.org/W6639703010","https://openalex.org/W6696798448","https://openalex.org/W6734062232","https://openalex.org/W6743755670","https://openalex.org/W6748163181","https://openalex.org/W6748224102","https://openalex.org/W6748565655","https://openalex.org/W6750410491"],"related_works":["https://openalex.org/W2084086280","https://openalex.org/W2027198738","https://openalex.org/W2397767885","https://openalex.org/W2062514649","https://openalex.org/W2083114070","https://openalex.org/W237421766","https://openalex.org/W2383111961","https://openalex.org/W2365952365","https://openalex.org/W3013712002","https://openalex.org/W2090725078"],"abstract_inverted_index":{"To":[0],"realize":[1],"the":[2,44,51,91,95,98,102,106,110],"promise":[3,24],"of":[4,16,46,77,93,97,109],"ubiquitous":[5],"embedded":[6],"deep":[7],"network":[8],"inference,":[9],"it":[10],"is":[11,88],"essential":[12],"to":[13,61,72,82,90,157,162],"seek":[14,70],"limits":[15],"energy":[17,26],"and":[18,27,37,69,104,125,132,139],"area":[19,28],"efficiency.":[20],"Low-precision":[21],"networks":[22,39,160],"offer":[23],"as":[25],"scale":[29],"down":[30],"quadratically":[31],"with":[32,66,120,150],"precision.":[33],"We":[34,55],"demonstrate":[35],"8-":[36],"4-bit":[38],"that":[40,57,146,171],"meet":[41],"or":[42],"exceed":[43],"accuracy":[45],"their":[47],"full-precision":[48],"versions":[49],"on":[50],"ImageNet":[52],"classification":[53],"benchmark.":[54],"hypothesize":[56],"gradient":[58,111],"noise":[59,127],"due":[60],"quantization":[62],"during":[63,134],"training":[64,86,135,137],"increases":[65],"reduced":[67],"precision,":[68],"ways":[71],"overcome":[73],"this.":[74],"The":[75],"number":[76],"iterations":[78],"required":[79],"by":[80,118,129,136],"SGD":[81],"achieve":[83],"a":[84],"given":[85],"error":[87],"related":[89],"square":[92],"(a)":[94],"distance":[96,117],"initial":[99],"solution":[100,116],"from":[101],"final":[103],"(b)":[105],"maximum":[107],"variance":[108],"estimates.":[112],"Accordingly,":[113],"we":[114],"reduce":[115],"starting":[119],"pretrained":[121],"fp32":[122,163],"baseline":[123,165],"networks,":[124],"combat":[126],"introduced":[128],"quantizing":[130],"weights":[131],"activations":[133],"longer":[138],"reducing":[140],"learning":[141],"rates.":[142],"Sensitivity":[143],"analysis":[144],"indicates":[145],"these":[147],"techniques,":[148],"coupled":[149],"activation":[151],"function":[152],"range":[153],"calibration,":[154],"are":[155],"sufficient":[156],"discover":[158],"low-precision":[159],"close":[161],"precision":[164],"networks.":[166],"Our":[167],"results":[168],"provide":[169],"evidence":[170],"4-bits":[172],"suffice":[173],"for":[174],"classification.":[175]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":6},{"year":2024,"cited_by_count":7},{"year":2023,"cited_by_count":5},{"year":2022,"cited_by_count":9},{"year":2021,"cited_by_count":7},{"year":2020,"cited_by_count":5}],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}