{"id":"https://openalex.org/W7154996881","doi":"https://doi.org/10.48550/arxiv.2604.16113","title":"Co-Design of CNN Accelerators for TinyML using Approximate Matrix Decomposition","display_name":"Co-Design of CNN Accelerators for TinyML using Approximate Matrix Decomposition","publication_year":2026,"publication_date":"2026-04-17","ids":{"openalex":"https://openalex.org/W7154996881","doi":"https://doi.org/10.48550/arxiv.2604.16113"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.16113","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.16113","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2604.16113","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5134010131","display_name":"Jos\u00e9 Juan Hern\u00e1ndez Morales","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Morales, Jos\u00e9 Juan Hern\u00e1ndez","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134058503","display_name":"Georgios Mentzos","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Mentzos, Georgios","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134068799","display_name":"Frank Hannig","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Hannig, Frank","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5006315507","display_name":"Konstantinos Balaskas","orcid":"https://orcid.org/0000-0003-2886-6896"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Balaskas, Konstantinos","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5051495057","display_name":"Georgios Zervakis","orcid":"https://orcid.org/0000-0001-8110-7122"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zervakis, Georgios","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5134026749","display_name":"J\u00f6rg Henkel","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Henkel, J\u00f6rg","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5134082293","display_name":"J\u00fcrgen Teich","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Teich, J\u00fcrgen","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9496999979019165,"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":0.9496999979019165,"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/T10904","display_name":"Embedded Systems Design Techniques","score":0.004800000227987766,"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"}},{"id":"https://openalex.org/T14347","display_name":"Big Data and Digital Economy","score":0.004399999976158142,"subfield":{"id":"https://openalex.org/subfields/1710","display_name":"Information Systems"},"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/inference","display_name":"Inference","score":0.7099000215530396},{"id":"https://openalex.org/keywords/latency","display_name":"Latency (audio)","score":0.6179999709129333},{"id":"https://openalex.org/keywords/implementation","display_name":"Implementation","score":0.5753999948501587},{"id":"https://openalex.org/keywords/field-programmable-gate-array","display_name":"Field-programmable gate array","score":0.560699999332428},{"id":"https://openalex.org/keywords/matrix-decomposition","display_name":"Matrix decomposition","score":0.4724000096321106},{"id":"https://openalex.org/keywords/optimization-problem","display_name":"Optimization problem","score":0.4108000099658966},{"id":"https://openalex.org/keywords/decomposition","display_name":"Decomposition","score":0.3984000086784363}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.772599995136261},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.7099000215530396},{"id":"https://openalex.org/C82876162","wikidata":"https://www.wikidata.org/wiki/Q17096504","display_name":"Latency (audio)","level":2,"score":0.6179999709129333},{"id":"https://openalex.org/C26713055","wikidata":"https://www.wikidata.org/wiki/Q245962","display_name":"Implementation","level":2,"score":0.5753999948501587},{"id":"https://openalex.org/C42935608","wikidata":"https://www.wikidata.org/wiki/Q190411","display_name":"Field-programmable gate array","level":2,"score":0.560699999332428},{"id":"https://openalex.org/C42355184","wikidata":"https://www.wikidata.org/wiki/Q1361088","display_name":"Matrix decomposition","level":3,"score":0.4724000096321106},{"id":"https://openalex.org/C137836250","wikidata":"https://www.wikidata.org/wiki/Q984063","display_name":"Optimization problem","level":2,"score":0.4108000099658966},{"id":"https://openalex.org/C124681953","wikidata":"https://www.wikidata.org/wiki/Q339062","display_name":"Decomposition","level":2,"score":0.3984000086784363},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.39259999990463257},{"id":"https://openalex.org/C17349429","wikidata":"https://www.wikidata.org/wiki/Q1049914","display_name":"Matrix multiplication","level":3,"score":0.38580000400543213},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3425999879837036},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.34049999713897705},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.337799996137619},{"id":"https://openalex.org/C206345919","wikidata":"https://www.wikidata.org/wiki/Q20380951","display_name":"Resource (disambiguation)","level":2,"score":0.3075999915599823},{"id":"https://openalex.org/C173608175","wikidata":"https://www.wikidata.org/wiki/Q232661","display_name":"Parallel computing","level":1,"score":0.299699991941452},{"id":"https://openalex.org/C106487976","wikidata":"https://www.wikidata.org/wiki/Q685816","display_name":"Matrix (chemical analysis)","level":2,"score":0.29030001163482666},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.28790000081062317},{"id":"https://openalex.org/C206729178","wikidata":"https://www.wikidata.org/wiki/Q2271896","display_name":"Scheduling (production processes)","level":2,"score":0.27300000190734863},{"id":"https://openalex.org/C58013763","wikidata":"https://www.wikidata.org/wiki/Q5754574","display_name":"High-level synthesis","level":3,"score":0.2531000077724457},{"id":"https://openalex.org/C46637626","wikidata":"https://www.wikidata.org/wiki/Q6693015","display_name":"Low latency (capital markets)","level":2,"score":0.25220000743865967}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.16113","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.16113","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2604.16113","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.16113","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"The":[0,21,193],"paradigm":[1],"shift":[2],"towards":[3],"local":[4],"and":[5,38,111,129,184,201,229],"on-device":[6],"inference":[7,52,140],"under":[8,34],"stringent":[9],"resource":[10,230],"constraints":[11,183],"is":[12,26,90],"represented":[13],"by":[14],"the":[15,42,64,108],"tiny":[16],"machine":[17],"learning":[18],"(TinyML)":[19],"domain.":[20],"primary":[22],"goal":[23],"of":[24,45,188,213,243],"TinyML":[25,67,215],"to":[27,50,107,116,154,170,176,181,246],"integrate":[28],"intelligence":[29],"into":[30],"tiny,":[31],"low-cost":[32],"devices":[33,47],"strict":[35,182,227],"resource,":[36],"energy,":[37],"latency":[39,60,77,236],"constraints.":[40],"However,":[41,142],"ultra-resource-constrained":[43],"nature":[44],"these":[46],"can":[48,56,112],"lead":[49,153],"increased":[51],"execution":[53],"time,":[54,66],"which":[55],"be":[57,113],"detrimental":[58],"in":[59,150,174,221],"critical":[61],"applications.":[62],"At":[63],"same":[65],"applications":[68],"are":[69,85,147],"often":[70],"associated":[71],"with":[72,122,238],"sensitive":[73],"data.":[74],"As":[75],"such,":[76],"optimization":[78,100],"approaches":[79,101,146],"that":[80,102,165,225],"rely":[81],"on":[82],"training":[83,109],"samples":[84],"infeasible":[86],"when":[87],"such":[88,126],"data":[89],"unavailable,":[91],"proprietary,":[92],"or":[93,190],"sensitive,":[94],"highlighting":[95],"a":[96,163,171],"pressing":[97],"need":[98,187],"for":[99,138,207],"do":[103],"not":[104],"require":[105],"access":[106],"dataset":[110],"applied":[114],"directly":[115],"pre-trained":[117],"models.":[118],"Replacing":[119],"costly":[120],"multiplications":[121],"more":[123],"hardware-efficient":[124],"operations,":[125],"as":[127,134],"shifts":[128],"additions,":[130],"has":[131],"been":[132],"proposed":[133],"an":[135,233,239],"effective":[136],"method":[137],"reducing":[139],"latency.":[141],"post-training":[143],"power-of-two":[144],"(Po2)":[145],"scarce":[148],"and,":[149],"many":[151],"cases,":[152],"unacceptable":[155],"accuracy":[156,228,241],"loss.":[157],"In":[158],"this":[159],"work,":[160],"we":[161],"propose":[162],"framework":[164,196],"applies":[166],"approximate":[167],"matrix":[168,199],"decomposition":[169],"given":[172],"CNN":[173,204],"order":[175],"optimize":[177],"hardware":[178],"implementations":[179,224],"subject":[180],"without":[185],"any":[186],"re-training":[189],"fine-tuning":[191],"steps.":[192],"genetic":[194],"algorithm-driven":[195],"explores":[197],"different":[198,214],"decompositions":[200],"resulting":[202],"multiplier-less":[203],"accelerator":[205],"designs":[206],"FPGA":[208,250],"targets.":[209],"A":[210],"comprehensive":[211],"evaluation":[212],"benchmarks":[216],"demonstrates":[217],"our":[218],"framework's":[219],"efficacy":[220],"generating":[222],"latency-optimized":[223],"satisfy":[226],"constraints,":[231],"achieving":[232],"average":[234,240],"33%":[235],"improvement":[237],"loss":[242],"1.3%":[244],"compared":[245],"typical":[247],"systolic":[248],"array-based":[249],"accelerators.":[251]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-04-21T00:00:00"}