{"id":"https://openalex.org/W4383501678","doi":"https://doi.org/10.1109/aicas57966.2023.10168577","title":"Free Bits: Latency Optimization of Mixed-Precision Quantized Neural Networks on the Edge","display_name":"Free Bits: Latency Optimization of Mixed-Precision Quantized Neural Networks on the Edge","publication_year":2023,"publication_date":"2023-06-11","ids":{"openalex":"https://openalex.org/W4383501678","doi":"https://doi.org/10.1109/aicas57966.2023.10168577"},"language":"en","primary_location":{"id":"doi:10.1109/aicas57966.2023.10168577","is_oa":false,"landing_page_url":"https://doi.org/10.1109/aicas57966.2023.10168577","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2023 IEEE 5th International Conference on Artificial Intelligence Circuits and Systems (AICAS)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://hdl.handle.net/11585/953199","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5007387832","display_name":"Georg Rutishauser","orcid":"https://orcid.org/0000-0001-8875-7611"},"institutions":[{"id":"https://openalex.org/I35440088","display_name":"ETH Zurich","ror":"https://ror.org/05a28rw58","country_code":"CH","type":"education","lineage":["https://openalex.org/I2799323385","https://openalex.org/I35440088"]}],"countries":["CH"],"is_corresponding":false,"raw_author_name":"Georg Rutishauser","raw_affiliation_strings":["ETH Zürich,Departement Informationstechnologie und Elektrotechnik,Switzerland"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ETH Zürich,Departement Informationstechnologie und Elektrotechnik,Switzerland","institution_ids":["https://openalex.org/I35440088"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5038717922","display_name":"Francesco Conti","orcid":"https://orcid.org/0000-0002-7924-933X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Francesco Conti","raw_affiliation_strings":["Università di Bologna,Dipartimento di Ingegneria Dell’Energia Elettrica e Dell’Informazione,Bologna,Italy"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Università di Bologna,Dipartimento di Ingegneria Dell’Energia Elettrica e Dell’Informazione,Bologna,Italy","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5043408422","display_name":"Luca Benini","orcid":"https://orcid.org/0000-0001-8068-3806"},"institutions":[{"id":"https://openalex.org/I35440088","display_name":"ETH Zurich","ror":"https://ror.org/05a28rw58","country_code":"CH","type":"education","lineage":["https://openalex.org/I2799323385","https://openalex.org/I35440088"]}],"countries":["CH"],"is_corresponding":false,"raw_author_name":"Luca Benini","raw_affiliation_strings":["ETH Zürich,Departement Informationstechnologie und Elektrotechnik,Switzerland"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ETH Zürich,Departement Informationstechnologie und Elektrotechnik,Switzerland","institution_ids":["https://openalex.org/I35440088"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.4254,"has_fulltext":false,"cited_by_count":4,"citation_normalized_percentile":{"value":0.61460086,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":97,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"5"},"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":0.9983000159263611,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.9955999851226807,"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":0.7023261785507202},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.6057892441749573},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.5793732404708862},{"id":"https://openalex.org/keywords/enhanced-data-rates-for-gsm-evolution","display_name":"Enhanced Data Rates for GSM Evolution","score":0.4565240740776062},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.18632209300994873},{"id":"https://openalex.org/keywords/telecommunications","display_name":"Telecommunications","score":0.10829439759254456}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7023261785507202},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.6057892441749573},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.5793732404708862},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.4565240740776062},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.18632209300994873},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.10829439759254456}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1109/aicas57966.2023.10168577","is_oa":false,"landing_page_url":"https://doi.org/10.1109/aicas57966.2023.10168577","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2023 IEEE 5th International Conference on Artificial Intelligence Circuits and Systems (AICAS)","raw_type":"proceedings-article"},{"id":"pmh:oai:cris.unibo.it:11585/953199","is_oa":true,"landing_page_url":"https://hdl.handle.net/11585/953199","pdf_url":null,"source":{"id":"https://openalex.org/S4306402579","display_name":"Archivio istituzionale della ricerca (Alma Mater Studiorum Universit\u00e0 di Bologna)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I4210117483","host_organization_name":"Istituto di Ematologia di Bologna","host_organization_lineage":["https://openalex.org/I4210117483"],"host_organization_lineage_names":[],"type":"repository"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"info:eu-repo/semantics/conferenceObject"}],"best_oa_location":{"id":"pmh:oai:cris.unibo.it:11585/953199","is_oa":true,"landing_page_url":"https://hdl.handle.net/11585/953199","pdf_url":null,"source":{"id":"https://openalex.org/S4306402579","display_name":"Archivio istituzionale della ricerca (Alma Mater Studiorum Universit\u00e0 di Bologna)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I4210117483","host_organization_name":"Istituto di Ematologia di Bologna","host_organization_lineage":["https://openalex.org/I4210117483"],"host_organization_lineage_names":[],"type":"repository"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"info:eu-repo/semantics/conferenceObject"},"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320334322","display_name":"HORIZON EUROPE Framework Programme","ror":null}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":20,"referenced_works":["https://openalex.org/W2117539524","https://openalex.org/W2963163009","https://openalex.org/W2963526839","https://openalex.org/W3004905356","https://openalex.org/W3024366773","https://openalex.org/W3034887213","https://openalex.org/W3049586678","https://openalex.org/W3137875211","https://openalex.org/W3208112639","https://openalex.org/W3214153136","https://openalex.org/W4212774754","https://openalex.org/W4297775537","https://openalex.org/W4311277703","https://openalex.org/W6734062232","https://openalex.org/W6762754082","https://openalex.org/W6767597771","https://openalex.org/W6773930524","https://openalex.org/W6779885597","https://openalex.org/W6784581357","https://openalex.org/W6803926790"],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W4396701345","https://openalex.org/W2376932109","https://openalex.org/W2001405890","https://openalex.org/W4396696052"],"abstract_inverted_index":{"Mixed-precision":[0],"quantization,":[1],"where":[2],"a":[3,45,51,58,65,75,93,120,125,174],"deep":[4],"neural":[5],"network\u2019s":[6],"layers":[7],"are":[8],"quantized":[9],"to":[10,16,69,107,115,137,165],"different":[11,101],"precisions,":[12],"offers":[13],"the":[14,18,37,89,129,158],"opportunity":[15],"optimize":[17],"trade-offs":[19],"between":[20],"model":[21,118],"size,":[22],"latency,":[23],"and":[24,85,87],"statistical":[25],"accuracy":[26,122,153],"beyond":[27],"what":[28],"can":[29],"be":[30],"achieved":[31],"with":[32,100,144,163],"homogeneous-bit-width":[33],"quantization.":[34],"To":[35],"navigate":[36],"in-tractable":[38],"search":[39,53,61,167],"space":[40],"of":[41,57,95,111,160],"mixed-precision":[42,71],"configurations":[43,72],"for":[44,74,148,176],"given":[46],"network,":[47],"this":[48],"paper":[49],"proposes":[50],"hybrid":[52],"methodology.":[54],"It":[55],"consists":[56],"hardware-agnostic":[59],"differentiable":[60,166],"algorithm":[62,82],"followed":[63],"by":[64],"hardware-aware":[66],"heuristic":[67],"optimization":[68],"find":[70],"latency-optimized":[73],"specific":[76],"hardware":[77,102,146],"target.":[78],"We":[79,104,133],"evaluate":[80],"our":[81,161],"on":[83,92,128,142],"MobileNetV1":[84],"MobileNetV2":[86],"deploy":[88],"resulting":[90],"networks":[91],"family":[94],"multi-core":[96],"RISC-V":[97],"microcontroller":[98],"platforms":[99],"characteristics.":[103],"achieve":[105],"up":[106],"28.6":[108],"%":[109],"reduction":[110],"end-to-end":[112],"latency":[113],"compared":[114],"an":[116,138],"8-bit":[117,139],"at":[119,151],"negligible":[121,152],"drop":[123],"from":[124],"full-precision":[126],"baseline":[127],"1000-class":[130],"ImageNet":[131],"dataset.":[132],"demonstrate":[134],"speedups":[135],"relative":[136],"baseline,":[140],"even":[141],"systems":[143],"no":[145],"support":[147],"sub-byte":[149],"arithmetic":[150],"drop.":[154],"Furthermore,":[155],"we":[156],"show":[157],"superiority":[159],"approach":[162],"respect":[164],"targeting":[168],"reduced":[169],"binary":[170],"operation":[171],"counts":[172],"as":[173],"proxy":[175],"latency.":[177]},"counts_by_year":[{"year":2024,"cited_by_count":4}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}