{"id":"https://openalex.org/W4379930100","doi":"https://doi.org/10.1109/access.2023.3284388","title":"Impact of Mixed Precision Techniques on Training and Inference Efficiency of Deep Neural Networks","display_name":"Impact of Mixed Precision Techniques on Training and Inference Efficiency of Deep Neural Networks","publication_year":2023,"publication_date":"2023-01-01","ids":{"openalex":"https://openalex.org/W4379930100","doi":"https://doi.org/10.1109/access.2023.3284388"},"language":"en","primary_location":{"id":"doi:10.1109/access.2023.3284388","is_oa":true,"landing_page_url":"https://doi.org/10.1109/access.2023.3284388","pdf_url":"https://ieeexplore.ieee.org/ielx7/6287639/10005208/10146255.pdf","source":{"id":"https://openalex.org/S2485537415","display_name":"IEEE Access","issn_l":"2169-3536","issn":["2169-3536"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Access","raw_type":"journal-article"},"type":"article","indexed_in":["crossref","doaj"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://ieeexplore.ieee.org/ielx7/6287639/10005208/10146255.pdf","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5000215860","display_name":"Marion D\u00f6rrich","orcid":"https://orcid.org/0000-0003-2956-1437"},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Marion D\u00f6rrich","raw_affiliation_strings":["Department of Artificial Intelligence in Biomedical Engineering, Friedrich-Alexander-Universi&#x00E4;t Erlangen-N&#x00FC;rnberg, Erlangen, Germany"],"affiliations":[{"raw_affiliation_string":"Department of Artificial Intelligence in Biomedical Engineering, Friedrich-Alexander-Universi&#x00E4;t Erlangen-N&#x00FC;rnberg, Erlangen, Germany","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101689144","display_name":"Mingcheng Fan","orcid":"https://orcid.org/0009-0001-8913-9847"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mingcheng Fan","raw_affiliation_strings":["Department of Artificial Intelligence in Biomedical Engineering, Friedrich-Alexander-Universi&#x00E4;t Erlangen-N&#x00FC;rnberg, Erlangen, Germany"],"affiliations":[{"raw_affiliation_string":"Department of Artificial Intelligence in Biomedical Engineering, Friedrich-Alexander-Universi&#x00E4;t Erlangen-N&#x00FC;rnberg, Erlangen, Germany","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5088428516","display_name":"Andreas M. Kist","orcid":"https://orcid.org/0000-0003-3643-7776"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Andreas M. Kist","raw_affiliation_strings":["Department of Artificial Intelligence in Biomedical Engineering, Friedrich-Alexander-Universi&#x00E4;t Erlangen-N&#x00FC;rnberg, Erlangen, Germany"],"affiliations":[{"raw_affiliation_string":"Department of Artificial Intelligence in Biomedical Engineering, Friedrich-Alexander-Universi&#x00E4;t Erlangen-N&#x00FC;rnberg, Erlangen, Germany","institution_ids":[]}]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5000215860"],"corresponding_institution_ids":[],"apc_list":{"value":1850,"currency":"USD","value_usd":1850},"apc_paid":{"value":1758,"currency":"EUR","value_usd":1896},"fwci":2.6381,"has_fulltext":true,"cited_by_count":22,"citation_normalized_percentile":{"value":0.91716054,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":90,"max":100},"biblio":{"volume":"11","issue":null,"first_page":"57627","last_page":"57634"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9991000294685364,"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":0.9991000294685364,"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/T11948","display_name":"Machine Learning in Materials Science","score":0.9681000113487244,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T12535","display_name":"Machine Learning and Data Classification","score":0.9563999772071838,"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":0.7984375357627869},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.6604714393615723},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.6363809108734131},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.6036726236343384},{"id":"https://openalex.org/keywords/efficient-energy-use","display_name":"Efficient energy use","score":0.6004034280776978},{"id":"https://openalex.org/keywords/speedup","display_name":"Speedup","score":0.5399112701416016},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.5198022127151489},{"id":"https://openalex.org/keywords/energy-consumption","display_name":"Energy consumption","score":0.480295866727829},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.4748696982860565},{"id":"https://openalex.org/keywords/enhanced-data-rates-for-gsm-evolution","display_name":"Enhanced Data Rates for GSM Evolution","score":0.45639416575431824},{"id":"https://openalex.org/keywords/excavator","display_name":"Excavator","score":0.41135793924331665},{"id":"https://openalex.org/keywords/parallel-computing","display_name":"Parallel computing","score":0.13636720180511475}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7984375357627869},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.6604714393615723},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6363809108734131},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6036726236343384},{"id":"https://openalex.org/C2742236","wikidata":"https://www.wikidata.org/wiki/Q924713","display_name":"Efficient energy use","level":2,"score":0.6004034280776978},{"id":"https://openalex.org/C68339613","wikidata":"https://www.wikidata.org/wiki/Q1549489","display_name":"Speedup","level":2,"score":0.5399112701416016},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5198022127151489},{"id":"https://openalex.org/C2780165032","wikidata":"https://www.wikidata.org/wiki/Q16869822","display_name":"Energy consumption","level":2,"score":0.480295866727829},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.4748696982860565},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.45639416575431824},{"id":"https://openalex.org/C141392084","wikidata":"https://www.wikidata.org/wiki/Q182661","display_name":"Excavator","level":2,"score":0.41135793924331665},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.13636720180511475},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"id":"https://openalex.org/C78519656","wikidata":"https://www.wikidata.org/wiki/Q101333","display_name":"Mechanical engineering","level":1,"score":0.0},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.0},{"id":"https://openalex.org/C18903297","wikidata":"https://www.wikidata.org/wiki/Q7150","display_name":"Ecology","level":1,"score":0.0},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1109/access.2023.3284388","is_oa":true,"landing_page_url":"https://doi.org/10.1109/access.2023.3284388","pdf_url":"https://ieeexplore.ieee.org/ielx7/6287639/10005208/10146255.pdf","source":{"id":"https://openalex.org/S2485537415","display_name":"IEEE Access","issn_l":"2169-3536","issn":["2169-3536"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Access","raw_type":"journal-article"},{"id":"pmh:oai:aleph.bib-bvb.de:BVB01-035238921","is_oa":true,"landing_page_url":"https://open.fau.de/handle/openfau/31615","pdf_url":"https://nbn-resolving.org/urn:nbn:de:101:1-2409101343121.952998720733","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"pmh:oai:doaj.org/article:61ee2838b8e443b1bbf6d76b1a4cfa37","is_oa":true,"landing_page_url":"https://doaj.org/article/61ee2838b8e443b1bbf6d76b1a4cfa37","pdf_url":null,"source":{"id":"https://openalex.org/S112646816","display_name":"SHILAP Revista de lepidopterolog\u00eda","issn_l":"0300-5267","issn":["0300-5267","2340-4078"],"is_oa":true,"is_in_doaj":true,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"IEEE Access, Vol 11, Pp 57627-57634 (2023)","raw_type":"article"}],"best_oa_location":{"id":"doi:10.1109/access.2023.3284388","is_oa":true,"landing_page_url":"https://doi.org/10.1109/access.2023.3284388","pdf_url":"https://ieeexplore.ieee.org/ielx7/6287639/10005208/10146255.pdf","source":{"id":"https://openalex.org/S2485537415","display_name":"IEEE Access","issn_l":"2169-3536","issn":["2169-3536"],"is_oa":true,"is_in_doaj":true,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE Access","raw_type":"journal-article"},"sustainable_development_goals":[{"display_name":"Affordable and clean energy","score":0.8999999761581421,"id":"https://metadata.un.org/sdg/7"}],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4379930100.pdf","grobid_xml":"https://content.openalex.org/works/W4379930100.grobid-xml"},"referenced_works_count":35,"referenced_works":["https://openalex.org/W1522301498","https://openalex.org/W1901129140","https://openalex.org/W2119144962","https://openalex.org/W2233116163","https://openalex.org/W2271840356","https://openalex.org/W2606722458","https://openalex.org/W2763421725","https://openalex.org/W2809624076","https://openalex.org/W2962914239","https://openalex.org/W2963809228","https://openalex.org/W3036483348","https://openalex.org/W3046665618","https://openalex.org/W3048300167","https://openalex.org/W3106250896","https://openalex.org/W3137147200","https://openalex.org/W3175856466","https://openalex.org/W3184320749","https://openalex.org/W3185946996","https://openalex.org/W3214897310","https://openalex.org/W4212896416","https://openalex.org/W4284664530","https://openalex.org/W4287204036","https://openalex.org/W4288346545","https://openalex.org/W4391904442","https://openalex.org/W6631190155","https://openalex.org/W6639824700","https://openalex.org/W6677580257","https://openalex.org/W6694517276","https://openalex.org/W6745245109","https://openalex.org/W6753069482","https://openalex.org/W6763653508","https://openalex.org/W6785652829","https://openalex.org/W6791940793","https://openalex.org/W6794426595","https://openalex.org/W6849346239"],"related_works":["https://openalex.org/W3003601017","https://openalex.org/W2038528908","https://openalex.org/W2997898942","https://openalex.org/W2997279469","https://openalex.org/W2901741596","https://openalex.org/W2363098002","https://openalex.org/W2347606691","https://openalex.org/W2466671794","https://openalex.org/W2097707447","https://openalex.org/W179829755"],"abstract_inverted_index":{"In":[0],"the":[1,33,69,78,95,144,155,162,182,207,215],"deep":[2,114],"learning":[3,115],"community,":[4],"increasingly":[5],"large":[6],"models":[7,156],"are":[8],"being":[9],"developed,":[10],"leading":[11],"to":[12,53,66,105,132,143,173,211],"rapidly":[13],"growing":[14],"computational":[15],"costs":[16],"and":[17,74,83,119,147,177,189,213,218],"energy":[18,34,55,70,187],"costs.":[19],"Recently,":[20],"a":[21,63,98,133,165,221],"new":[22],"trend":[23],"has":[24,45,220],"been":[25],"arising,":[26],"advocating":[27],"that":[28,47,125],"researchers":[29,172],"should":[30],"also":[31],"report":[32],"efficiency":[35,71],"besides":[36],"their":[37,41,175,195],"model\u2019s":[38,183],"performance":[39],"in":[40,205],"papers.":[42],"Previous":[43],"research":[44],"shown":[46],"reduced":[48],"precision":[49,82,128],"can":[50,129],"be":[51],"helpful":[52,204],"improve":[54,68],"efficiency.":[56],"Based":[57],"on":[58,86,152,157,224],"this":[59],"finding,":[60],"we":[61],"propose":[62],"simple":[64],"practice":[65],"effectively":[67],"of":[72,97,167],"training":[73,77,176],"inference,":[75],"i.e.,":[76],"model":[79,117,196],"with":[80,103],"mixed":[81,127],"deploying":[84],"it":[85],"Edge":[87,158],"TPUs.":[88],"We":[89],"evaluated":[90],"its":[91],"effectiveness":[92],"by":[93,164],"comparing":[94],"speed-up":[96],"state-of-the-art":[99],"semantic":[100],"segmentation":[101],"architecture":[102,197],"respect":[104],"different":[106,112],"typical":[107],"usage":[108],"scenarios,":[109],"including":[110],"using":[111],"devices,":[113],"frameworks,":[116],"sizes,":[118],"batch":[120],"sizes.":[121],"Our":[122,169],"results":[123],"show":[124],"enabled":[126],"gain":[130],"up":[131],"<inline-formula":[134],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[135],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">":[136],"<tex-math":[137],"notation=\"LaTeX\">$1.9\\times":[138],"$":[139],"</tex-math></inline-formula>":[140],"speedup":[141],"compared":[142],"most":[145],"common":[146],"default":[148],"float32":[149],"data":[150],"type":[151],"GPUs.":[153],"Deploying":[154],"TPU":[159],"further":[160],"boosted":[161],"inference":[163,178],"factor":[166],"6.":[168],"approach":[170,202],"allows":[171],"accelerate":[174],"procedures":[179],"without":[180,193],"jeopardizing":[181],"accuracy,":[184],"meanwhile":[185],"reducing":[186,206],"consumption":[188],"electricity":[190],"cost":[191],"easily":[192],"changing":[194],"or":[198],"retraining.":[199],"Furthermore,":[200],"our":[201],"is":[203],"carbon":[208],"footprint":[209],"used":[210],"train":[212],"deploy":[214],"neural":[216],"network":[217],"thus":[219],"positive":[222],"effect":[223],"environmental":[225],"resources.":[226]},"counts_by_year":[{"year":2026,"cited_by_count":2},{"year":2025,"cited_by_count":13},{"year":2024,"cited_by_count":6},{"year":2023,"cited_by_count":1}],"updated_date":"2026-04-04T16:13:02.066488","created_date":"2025-10-10T00:00:00"}