{"id":"https://openalex.org/W7127069821","doi":"https://doi.org/10.1145/3787256.3787272","title":"A High-Performance ReLU-Based Hardware Accelerator with Simplified Quantization for Efficient Vision Transformers on Edge Devices","display_name":"A High-Performance ReLU-Based Hardware Accelerator with Simplified Quantization for Efficient Vision Transformers on Edge Devices","publication_year":2025,"publication_date":"2025-11-21","ids":{"openalex":"https://openalex.org/W7127069821","doi":"https://doi.org/10.1145/3787256.3787272"},"language":null,"primary_location":{"id":"doi:10.1145/3787256.3787272","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3787256.3787272","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2025 8th International Conference on Computational Intelligence and Intelligent Systems","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3787256.3787272","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Yu-Wen Peng","orcid":"https://orcid.org/0009-0001-9676-0922"},"institutions":[{"id":"https://openalex.org/I118292597","display_name":"National Taipei University of Technology","ror":"https://ror.org/00cn92c09","country_code":"TW","type":"education","lineage":["https://openalex.org/I118292597"]}],"countries":["TW"],"is_corresponding":true,"raw_author_name":"Yu-Wen Peng","raw_affiliation_strings":["National Taipei University of Technology, Taipei, Taiwan,"],"raw_orcid":"https://orcid.org/0009-0001-9676-0922","affiliations":[{"raw_affiliation_string":"National Taipei University of Technology, Taipei, Taiwan,","institution_ids":["https://openalex.org/I118292597"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5017761295","display_name":"Pin-Chieh Hsieh","orcid":null},"institutions":[{"id":"https://openalex.org/I118292597","display_name":"National Taipei University of Technology","ror":"https://ror.org/00cn92c09","country_code":"TW","type":"education","lineage":["https://openalex.org/I118292597"]}],"countries":["TW"],"is_corresponding":false,"raw_author_name":"Pin-Chieh Hsieh","raw_affiliation_strings":["National Taipei University of Technology, Taipei, Taiwan,"],"raw_orcid":"https://orcid.org/0009-0004-0913-1987","affiliations":[{"raw_affiliation_string":"National Taipei University of Technology, Taipei, Taiwan,","institution_ids":["https://openalex.org/I118292597"]}]},{"author_position":"last","author":{"id":null,"display_name":"Yu-Cheng Fan","orcid":"https://orcid.org/0000-0002-9599-6415"},"institutions":[{"id":"https://openalex.org/I118292597","display_name":"National Taipei University of Technology","ror":"https://ror.org/00cn92c09","country_code":"TW","type":"education","lineage":["https://openalex.org/I118292597"]}],"countries":["TW"],"is_corresponding":false,"raw_author_name":"Yu-Cheng Fan","raw_affiliation_strings":["National Taipei University of Technology, Taipei, Taiwan,"],"raw_orcid":"https://orcid.org/0000-0002-9599-6415","affiliations":[{"raw_affiliation_string":"National Taipei University of Technology, Taipei, Taiwan,","institution_ids":["https://openalex.org/I118292597"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I118292597"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.60296377,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"107","last_page":"112"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11992","display_name":"CCD and CMOS Imaging Sensors","score":0.8830000162124634,"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"}},"topics":[{"id":"https://openalex.org/T11992","display_name":"CCD and CMOS Imaging Sensors","score":0.8830000162124634,"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":0.030799999833106995,"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/T10904","display_name":"Embedded Systems Design Techniques","score":0.008100000210106373,"subfield":{"id":"https://openalex.org/subfields/1708","display_name":"Hardware and Architecture"},"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/quantization","display_name":"Quantization (signal processing)","score":0.5928999781608582},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.5594000220298767},{"id":"https://openalex.org/keywords/edge-device","display_name":"Edge device","score":0.5587999820709229},{"id":"https://openalex.org/keywords/chip","display_name":"Chip","score":0.486299991607666},{"id":"https://openalex.org/keywords/power-consumption","display_name":"Power consumption","score":0.47369998693466187},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.400299996137619},{"id":"https://openalex.org/keywords/hardware-acceleration","display_name":"Hardware acceleration","score":0.38929998874664307},{"id":"https://openalex.org/keywords/process","display_name":"Process (computing)","score":0.3824000060558319}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6985999941825867},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.5928999781608582},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.5594000220298767},{"id":"https://openalex.org/C138236772","wikidata":"https://www.wikidata.org/wiki/Q25098575","display_name":"Edge device","level":3,"score":0.5587999820709229},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.5266000032424927},{"id":"https://openalex.org/C165005293","wikidata":"https://www.wikidata.org/wiki/Q1074500","display_name":"Chip","level":2,"score":0.486299991607666},{"id":"https://openalex.org/C2984118289","wikidata":"https://www.wikidata.org/wiki/Q29954","display_name":"Power consumption","level":3,"score":0.47369998693466187},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.400299996137619},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.3962000012397766},{"id":"https://openalex.org/C13164978","wikidata":"https://www.wikidata.org/wiki/Q600158","display_name":"Hardware acceleration","level":3,"score":0.38929998874664307},{"id":"https://openalex.org/C98045186","wikidata":"https://www.wikidata.org/wiki/Q205663","display_name":"Process (computing)","level":2,"score":0.3824000060558319},{"id":"https://openalex.org/C24326235","wikidata":"https://www.wikidata.org/wiki/Q126095","display_name":"Electronic engineering","level":1,"score":0.3605000078678131},{"id":"https://openalex.org/C2778456923","wikidata":"https://www.wikidata.org/wiki/Q5337692","display_name":"Edge computing","level":3,"score":0.3490000069141388},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.33889999985694885},{"id":"https://openalex.org/C162307627","wikidata":"https://www.wikidata.org/wiki/Q204833","display_name":"Enhanced Data Rates for GSM Evolution","level":2,"score":0.3231000006198883},{"id":"https://openalex.org/C2777480716","wikidata":"https://www.wikidata.org/wiki/Q23582796","display_name":"Resource consumption","level":2,"score":0.3061000108718872},{"id":"https://openalex.org/C9417928","wikidata":"https://www.wikidata.org/wiki/Q1070689","display_name":"Image processing","level":3,"score":0.30250000953674316},{"id":"https://openalex.org/C2742236","wikidata":"https://www.wikidata.org/wiki/Q924713","display_name":"Efficient energy use","level":2,"score":0.30090001225471497},{"id":"https://openalex.org/C2780165032","wikidata":"https://www.wikidata.org/wiki/Q16869822","display_name":"Energy consumption","level":2,"score":0.2992999851703644},{"id":"https://openalex.org/C118993495","wikidata":"https://www.wikidata.org/wiki/Q5042828","display_name":"Electrical efficiency","level":3,"score":0.29829999804496765},{"id":"https://openalex.org/C157764524","wikidata":"https://www.wikidata.org/wiki/Q1383412","display_name":"Throughput","level":3,"score":0.2840000092983246},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.2768000066280365},{"id":"https://openalex.org/C118021083","wikidata":"https://www.wikidata.org/wiki/Q610398","display_name":"System on a chip","level":2,"score":0.27059999108314514},{"id":"https://openalex.org/C104267543","wikidata":"https://www.wikidata.org/wiki/Q208163","display_name":"Signal processing","level":3,"score":0.25929999351501465}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1145/3787256.3787272","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3787256.3787272","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2025 8th International Conference on Computational Intelligence and Intelligent Systems","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.1145/3787256.3787272","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3787256.3787272","pdf_url":null,"source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2025 8th International Conference on Computational Intelligence and Intelligent Systems","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":11,"referenced_works":["https://openalex.org/W2112796928","https://openalex.org/W2169471437","https://openalex.org/W2618530766","https://openalex.org/W2949650786","https://openalex.org/W2962835968","https://openalex.org/W2963122961","https://openalex.org/W3094502228","https://openalex.org/W4321637298","https://openalex.org/W4376312661","https://openalex.org/W4385245566","https://openalex.org/W4406387987"],"related_works":[],"abstract_inverted_index":{"Vision":[0],"Transformer":[1],"(ViT)":[2],"models":[3,16],"demonstrate":[4],"substantial":[5],"promise":[6],"in":[7,26,96],"image":[8],"processing":[9],"tasks,":[10],"but":[11],"deploying":[12],"these":[13],"computationally":[14],"intensive":[15],"on":[17,49],"edge":[18,148],"devices":[19],"poses":[20],"significant":[21],"challenges":[22],"due":[23],"to":[24,44,79],"constraints":[25],"power":[27],"consumption":[28],"and":[29,74,114,137],"computational":[30],"resources.":[31],"This":[32],"paper":[33],"proposes":[34],"a":[35,75,103,116,141],"dedicated":[36],"hardware":[37,62],"accelerator":[38],"for":[39,71,147],"the":[40,109],"lightweight":[41],"EfficientViT":[42],"model":[43],"address":[45],"this":[46],"issue,":[47],"focusing":[48],"its":[50],"ReLU-based":[51],"global":[52],"attention":[53],"mechanism.":[54],"Our":[55],"approach":[56],"is":[57,94],"underpinned":[58],"by":[59],"two":[60],"key":[61],"optimization":[63],"strategies:":[64],"block-wise":[65],"attention,":[66],"which":[67],"decomposes":[68],"large":[69],"matrices":[70],"efficient":[72,85],"processing,":[73],"simplified":[76],"quantization":[77],"technique":[78],"convert":[80],"costly":[81],"floating-point":[82],"operations":[83],"into":[84],"integers":[86],"with":[87,122],"minimal":[88],"precision":[89],"loss.":[90],"The":[91],"proposed":[92],"design":[93],"fabricated":[95],"TSMC":[97],"40nm":[98],"process":[99],"technology.":[100],"Operating":[101],"at":[102],"maximum":[104],"frequency":[105],"of":[106,119,126],"1.29":[107],"GHz,":[108],"chip":[110],"consumes":[111],"340":[112],"mW":[113],"achieves":[115],"peak":[117],"throughput":[118],"368.9":[120],"GOPS,":[121],"an":[123,132],"area":[124],"efficiency":[125],"265.4":[127],"GOPS/mm\u00b2.":[128],"These":[129],"results":[130],"showcase":[131],"exceptional":[133],"balance":[134],"between":[135],"performance":[136],"resource":[138],"utilization,":[139],"delivering":[140],"high-performance":[142],"ViT":[143],"solution":[144],"ideally":[145],"suited":[146],"computing.":[149]},"counts_by_year":[],"updated_date":"2026-02-06T02:01:19.302388","created_date":"2026-02-03T00:00:00"}