{"id":"https://openalex.org/W7225832949","doi":"https://doi.org/10.48550/arxiv.2611.13641","title":"MOAI: Module-Optimizing Architecture for non-interactive secure transformer Inference","display_name":"MOAI: Module-Optimizing Architecture for non-interactive secure transformer Inference","publication_year":2026,"publication_date":"2026-11-10","ids":{"openalex":"https://openalex.org/W7225832949","doi":"https://doi.org/10.48550/arxiv.2611.13641"},"language":"en","primary_location":{"id":"doi:10.48550/arxiv.2611.13641","doi":"https://doi.org/10.48550/arxiv.2611.13641","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.13641","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,"listed_in":[],"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"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2611.13641","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5023093685","display_name":"Linru Zhang","orcid":"https://orcid.org/0009-0001-2645-4479","observed_orcids":["https://orcid.org/0009-0001-2645-4479"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhang, Linru","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101859738","display_name":"Xiangning Wang","orcid":"https://orcid.org/0009-0000-5742-3678","observed_orcids":["https://orcid.org/0009-0000-5742-3678"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Xiangning","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5065823811","display_name":"Jun Jie Sim","orcid":"https://orcid.org/0000-0002-3709-6929","observed_orcids":["https://orcid.org/0000-0002-3709-6929"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sim, Jun Jie","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161445323","display_name":"Zhicong Huang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Zhicong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161421445","display_name":"Jiahao Zhong","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhong, Jiahao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161488436","display_name":"Huaxiong Wang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Huaxiong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5161485719","display_name":"Pu Duan","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Duan, Pu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5125553369","display_name":"Kwok-Yan Lam","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lam, Kwok-Yan","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/T10237","display_name":"Cryptography and Data Security","score":0.9814000129699707,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T10237","display_name":"Cryptography and Data Security","score":0.9814000129699707,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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/T11424","display_name":"Security and Verification in Computing","score":0.006899999920278788,"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/T10951","display_name":"Cryptographic Implementations and Security","score":0.0015999999595806003,"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":[],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6516000032424927},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.5618000030517578},{"id":"https://openalex.org/C158338273","wikidata":"https://www.wikidata.org/wiki/Q2154943","display_name":"Homomorphic encryption","level":3,"score":0.5371000170707703},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4927999973297119},{"id":"https://openalex.org/C148730421","wikidata":"https://www.wikidata.org/wiki/Q141090","display_name":"Encryption","level":2,"score":0.4887999892234802},{"id":"https://openalex.org/C188441871","wikidata":"https://www.wikidata.org/wiki/Q7554146","display_name":"Softmax function","level":3,"score":0.4717000126838684},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.43709999322891235},{"id":"https://openalex.org/C79974875","wikidata":"https://www.wikidata.org/wiki/Q483639","display_name":"Cloud computing","level":2,"score":0.3928999900817871},{"id":"https://openalex.org/C86251818","wikidata":"https://www.wikidata.org/wiki/Q816754","display_name":"Benchmarking","level":2,"score":0.3765999972820282},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.361299991607666},{"id":"https://openalex.org/C17349429","wikidata":"https://www.wikidata.org/wiki/Q1049914","display_name":"Matrix multiplication","level":3,"score":0.34060001373291016},{"id":"https://openalex.org/C107598950","wikidata":"https://www.wikidata.org/wiki/Q259864","display_name":"Microarchitecture","level":2,"score":0.33880001306533813},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.3384999930858612},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.3312000036239624},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.31869998574256897},{"id":"https://openalex.org/C149635348","wikidata":"https://www.wikidata.org/wiki/Q193040","display_name":"Embedded system","level":1,"score":0.3084999918937683},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.3068999946117401},{"id":"https://openalex.org/C130367717","wikidata":"https://www.wikidata.org/wiki/Q189791","display_name":"Diagonal","level":2,"score":0.3057999908924103},{"id":"https://openalex.org/C1893757","wikidata":"https://www.wikidata.org/wiki/Q3653001","display_name":"Inversion (geology)","level":3,"score":0.29980000853538513},{"id":"https://openalex.org/C79403827","wikidata":"https://www.wikidata.org/wiki/Q3988","display_name":"Real-time computing","level":1,"score":0.29499998688697815},{"id":"https://openalex.org/C179518139","wikidata":"https://www.wikidata.org/wiki/Q5140297","display_name":"Coding (social sciences)","level":2,"score":0.28279998898506165},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.28209999203681946},{"id":"https://openalex.org/C134765980","wikidata":"https://www.wikidata.org/wiki/Q879126","display_name":"Bitwise operation","level":2,"score":0.27900001406669617},{"id":"https://openalex.org/C43364308","wikidata":"https://www.wikidata.org/wiki/Q8799","display_name":"Byte","level":2,"score":0.27720001339912415},{"id":"https://openalex.org/C6295992","wikidata":"https://www.wikidata.org/wiki/Q976521","display_name":"Cryptosystem","level":3,"score":0.2653999924659729},{"id":"https://openalex.org/C2989134064","wikidata":"https://www.wikidata.org/wiki/Q288510","display_name":"Execution time","level":2,"score":0.2646999955177307}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2611.13641","doi":"https://doi.org/10.48550/arxiv.2611.13641","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.13641","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,"listed_in":[],"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"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2611.13641","doi":"https://doi.org/10.48550/arxiv.2611.13641","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.13641","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,"listed_in":[],"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":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Privacy":[0],"concerns":[1],"have":[2],"been":[3],"raised":[4],"in":[5,15,109,117,128,160,192,211,241],"Large":[6],"Language":[7],"Models":[8],"(LLM)":[9],"inference":[10,30],"when":[11],"models":[12],"are":[13],"deployed":[14],"Cloud":[16],"Service":[17],"Providers":[18],"(CSP).":[19],"Homomorphic":[20],"encryption":[21],"(HE)":[22],"offers":[23],"a":[24,43,69,152,156,188,230],"promising":[25],"solution":[26],"by":[27,238],"enabling":[28],"secure":[29,59,169],"directly":[31],"over":[32],"encrypted":[33],"inputs.":[34],"However,":[35],"the":[36,99,126,142,165,183,236],"high":[37],"computational":[38],"overhead":[39],"of":[40,101,149,224],"HE":[41,103,107,138],"remains":[42],"major":[44],"bottleneck.":[45],"To":[46],"address":[47],"this":[48],"challenge,":[49],"we":[50,154],"propose":[51],"MOAI,":[52],"an":[53,221],"efficient":[54],"HE-based,":[55],"non-interactive":[56,168,216],"framework":[57,185],"for":[58,92],"transformer":[60,170,212,217],"inference.":[61,111,218],"MOAI":[62,132,181,240],"gains":[63],"significant":[64],"efficiency":[65],"improvement":[66],"from:":[67],"(1)":[68],"novel":[70],"evaluation":[71,161,193],"flow":[72],"that":[73,96],"combines":[74],"column":[75],"and":[76,94,121,186,206,213],"diagonal":[77],"packing":[78,114],"with":[79,208],"consistent":[80],"strategies":[81],"across":[82,145],"all":[83,146],"layers,":[84],"eliminating":[85],"expensive":[86],"format":[87],"conversions.":[88],"(2)":[89],"rotation-free":[90],"algorithms":[91],"Softmax":[93,205],"LayerNorm":[95,207],"significantly":[97],"reduce":[98],"number":[100],"costly":[102],"rotations,":[104],"removing":[105],"2,448":[106],"rotations":[108,116,127,139],"BERT-base":[110],"(3)":[112],"Column":[113],"removes":[115],"plaintext-ciphertext":[118],"matrix":[119,130,147],"multiplications":[120,148],"interleaved":[122],"batching":[123],"further":[124],"reduces":[125],"ciphertext-ciphertext":[129],"multiplications.":[131],"uses":[133],"at":[134],"least":[135],"1.7x":[136],"fewer":[137],"compared":[140,163,195],"to":[141,164,196],"state-of-the-art":[143,166],"works":[144],"BERT-base.":[150],"As":[151],"result,":[153],"achieve":[155,187],"$52.8":[157],"\\%$":[158,190],"reduction":[159,191],"time":[162,194,223],"HE-based":[167,215],"inference,":[171],"THOR":[172],"(Moon":[173],"et":[174,199],"al.,":[175,200],"CCS":[176],"25).":[177],"We":[178,219,234],"then":[179],"apply":[180],"on":[182,229],"Powerformer's":[184],"$55.7":[189],"Powerformer":[197],"(Park":[198],"ACL":[201],"25),":[202],"which":[203],"approximates":[204],"simpler":[209],"functions":[210],"proposes":[214],"report":[220],"amortized":[222],"2.36":[225],"minutes":[226],"per":[227],"input":[228],"single":[231],"GPU":[232],"environment.":[233],"show":[235],"extendibility":[237],"applying":[239],"LLaMA-3-8B.":[242],"Our":[243],"implementation":[244],"is":[245],"publicly":[246],"available":[247],"as":[248],"open":[249],"source.":[250]},"counts_by_year":[],"updated_date":"2026-11-12T12:28:57.059296","created_date":"2026-11-12T00:00:00"}