{"id":"https://openalex.org/W4388887209","doi":"https://doi.org/10.1145/3605764.3623902","title":"Equivariant Differentially Private Deep Learning: Why DP-SGD Needs Sparser Models","display_name":"Equivariant Differentially Private Deep Learning: Why DP-SGD Needs Sparser Models","publication_year":2023,"publication_date":"2023-11-21","ids":{"openalex":"https://openalex.org/W4388887209","doi":"https://doi.org/10.1145/3605764.3623902"},"language":"en","primary_location":{"id":"doi:10.1145/3605764.3623902","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3605764.3623902","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 16th ACM Workshop on Artificial Intelligence and Security","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3605764.3623902","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5072127418","display_name":"Florian A. H\u00f6lzl","orcid":"https://orcid.org/0000-0003-3218-7976"},"institutions":[{"id":"https://openalex.org/I62916508","display_name":"Technical University of Munich","ror":"https://ror.org/02kkvpp62","country_code":"DE","type":"education","lineage":["https://openalex.org/I62916508"]}],"countries":["DE"],"is_corresponding":true,"raw_author_name":"Florian A. H\u00f6lzl","raw_affiliation_strings":["Technical University of Munich, Munich, Germany"],"raw_orcid":"https://orcid.org/0000-0003-3218-7976","affiliations":[{"raw_affiliation_string":"Technical University of Munich, Munich, Germany","institution_ids":["https://openalex.org/I62916508"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5006461848","display_name":"Daniel Rueckert","orcid":"https://orcid.org/0000-0002-5683-5889"},"institutions":[{"id":"https://openalex.org/I62916508","display_name":"Technical University of Munich","ror":"https://ror.org/02kkvpp62","country_code":"DE","type":"education","lineage":["https://openalex.org/I62916508"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Daniel Rueckert","raw_affiliation_strings":["Imperial College London & Technical University of Munich, Munich, Germany"],"raw_orcid":"https://orcid.org/0000-0002-5683-5889","affiliations":[{"raw_affiliation_string":"Imperial College London & Technical University of Munich, Munich, Germany","institution_ids":["https://openalex.org/I62916508"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5000554104","display_name":"Georgios Kaissis","orcid":"https://orcid.org/0000-0001-8382-8062"},"institutions":[{"id":"https://openalex.org/I3018134672","display_name":"Helmholtz Zentrum M\u00fcnchen","ror":"https://ror.org/00cfam450","country_code":"DE","type":"facility","lineage":["https://openalex.org/I1305996414","https://openalex.org/I3018134672"]},{"id":"https://openalex.org/I62916508","display_name":"Technical University of Munich","ror":"https://ror.org/02kkvpp62","country_code":"DE","type":"education","lineage":["https://openalex.org/I62916508"]}],"countries":["DE"],"is_corresponding":false,"raw_author_name":"Georgios Kaissis","raw_affiliation_strings":["Helmholtz Munich & Technical University of Munich, Munich, Germany"],"raw_orcid":"https://orcid.org/0000-0001-8382-8062","affiliations":[{"raw_affiliation_string":"Helmholtz Munich & Technical University of Munich, Munich, Germany","institution_ids":["https://openalex.org/I3018134672","https://openalex.org/I62916508"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":["https://openalex.org/A5072127418"],"corresponding_institution_ids":["https://openalex.org/I62916508"],"apc_list":null,"apc_paid":null,"fwci":0.1704,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.58330311,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":91,"max":95},"biblio":{"volume":null,"issue":null,"first_page":"11","last_page":"22"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10764","display_name":"Privacy-Preserving Technologies in Data","score":1.0,"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"}},"topics":[{"id":"https://openalex.org/T10764","display_name":"Privacy-Preserving Technologies in Data","score":1.0,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9976999759674072,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9865999817848206,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.8078340291976929},{"id":"https://openalex.org/keywords/stochastic-gradient-descent","display_name":"Stochastic gradient descent","score":0.6465635299682617},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.6150498986244202},{"id":"https://openalex.org/keywords/differential-privacy","display_name":"Differential privacy","score":0.5904932618141174},{"id":"https://openalex.org/keywords/deep-learning","display_name":"Deep learning","score":0.5729839205741882},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.5407317876815796},{"id":"https://openalex.org/keywords/rendering","display_name":"Rendering (computer graphics)","score":0.46317559480667114},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.45798125863075256},{"id":"https://openalex.org/keywords/overhead","display_name":"Overhead (engineering)","score":0.44434720277786255},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.4195975661277771},{"id":"https://openalex.org/keywords/convolutional-neural-network","display_name":"Convolutional neural network","score":0.4110995829105377},{"id":"https://openalex.org/keywords/distributed-computing","display_name":"Distributed computing","score":0.35924428701400757},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.2520146369934082},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.22619470953941345}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8078340291976929},{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.6465635299682617},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6150498986244202},{"id":"https://openalex.org/C23130292","wikidata":"https://www.wikidata.org/wiki/Q5275358","display_name":"Differential privacy","level":2,"score":0.5904932618141174},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.5729839205741882},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5407317876815796},{"id":"https://openalex.org/C205711294","wikidata":"https://www.wikidata.org/wiki/Q176953","display_name":"Rendering (computer graphics)","level":2,"score":0.46317559480667114},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.45798125863075256},{"id":"https://openalex.org/C2779960059","wikidata":"https://www.wikidata.org/wiki/Q7113681","display_name":"Overhead (engineering)","level":2,"score":0.44434720277786255},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.4195975661277771},{"id":"https://openalex.org/C81363708","wikidata":"https://www.wikidata.org/wiki/Q17084460","display_name":"Convolutional neural network","level":2,"score":0.4110995829105377},{"id":"https://openalex.org/C120314980","wikidata":"https://www.wikidata.org/wiki/Q180634","display_name":"Distributed computing","level":1,"score":0.35924428701400757},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.2520146369934082},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.22619470953941345},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1145/3605764.3623902","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3605764.3623902","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 16th ACM Workshop on Artificial Intelligence and Security","raw_type":"proceedings-article"},{"id":"pmh:oai:push-zb.helmholtz-munich.de:69089","is_oa":false,"landing_page_url":"https://push-zb.helmholtz-munich.de/frontdoor.php?source_opus=69089","pdf_url":null,"source":{"id":"https://openalex.org/S7407055352","display_name":"PuSH - Publication Server of Helmholtz Zentrum M\u00fcnchen","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"In:. 2023. 11-22","raw_type":"info:eu-repo/semantics/conferenceObject"}],"best_oa_location":{"id":"doi:10.1145/3605764.3623902","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3605764.3623902","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 16th ACM Workshop on Artificial Intelligence and Security","raw_type":"proceedings-article"},"sustainable_development_goals":[{"score":0.4699999988079071,"id":"https://metadata.un.org/sdg/9","display_name":"Industry, innovation and infrastructure"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":14,"referenced_works":["https://openalex.org/W2073241381","https://openalex.org/W2082841340","https://openalex.org/W2086161653","https://openalex.org/W2154332973","https://openalex.org/W2473418344","https://openalex.org/W2990138404","https://openalex.org/W2991391304","https://openalex.org/W3035452548","https://openalex.org/W3129093240","https://openalex.org/W3212532487","https://openalex.org/W4205228770","https://openalex.org/W4286588524","https://openalex.org/W4287363917","https://openalex.org/W4288068240"],"related_works":["https://openalex.org/W3038283795","https://openalex.org/W2604501336","https://openalex.org/W2734500670","https://openalex.org/W2558166297","https://openalex.org/W2315671126","https://openalex.org/W798507144","https://openalex.org/W2964481303","https://openalex.org/W1751413323","https://openalex.org/W4401550921","https://openalex.org/W4361791424"],"abstract_inverted_index":{"Differentially":[0],"Private":[1],"Stochastic":[2],"Gradient":[3],"Descent":[4],"(DP-SGD)":[5],"limits":[6],"the":[7,28,87,103,183,208],"amount":[8],"of":[9,90,105,175,203],"private":[10,212],"information":[11],"deep":[12,215],"learning":[13,48,216],"models":[14,45],"can":[15,160],"memorize":[16],"during":[17],"training.":[18],"This":[19],"is":[20],"achieved":[21],"by":[22,136,138,186],"clipping":[23],"and":[24,31,78,92,120,156,206,213],"adding":[25],"noise":[26],"to":[27,109,177],"model's":[29],"gradients,":[30],"thus":[32,124],"networks":[33,142],"with":[34,53,115,146,164],"more":[35,58,187],"parameters":[36,205],"require":[37],"proportionally":[38],"stronger":[39],"perturbation.":[40],"As":[41],"a":[42,127,193],"result,":[43],"large":[44,79],"have":[46],"difficulties":[47],"useful":[49],"information,":[50],"rendering":[51],"training":[52,60,71,145],"DP-SGD":[54,91],"exceedingly":[55],"difficult":[56],"on":[57,66],"challenging":[59],"tasks.":[61],"Recent":[62],"research":[63],"has":[64],"focused":[65],"combating":[67],"this":[68,98],"challenge":[69],"through":[70],"adaptations":[72],"such":[73,112,134],"as":[74,126],"heavy":[75],"data":[76],"augmentation":[77],"batch":[80],"sizes.":[81],"However,":[82],"these":[83],"techniques":[84],"further":[85],"increase":[86,174],"computational":[88,167],"overhead":[89],"reduce":[93],"its":[94],"practical":[95],"applicability.":[96],"In":[97],"work,":[99],"we":[100,152,171],"propose":[101],"using":[102],"principle":[104],"sparse":[106],"model":[107,144,197],"design":[108,137,159],"solve":[110],"precisely":[111],"complex":[113],"tasks":[114],"fewer":[116],"parameters,":[117],"higher":[118],"accuracy,":[119],"in":[121,179],"less":[122],"time,":[123],"serving":[125],"promising":[128],"direction":[129],"for":[130,143,217],"DP-SGD.":[131],"We":[132],"achieve":[133,172],"sparsity":[135],"introducing":[139],"equivariant":[140,150],"convolutional":[141],"Differential":[147],"Privacy.":[148],"Using":[149],"networks,":[151],"show":[153],"that":[154,199],"small":[155],"efficient":[157,196],"architecture":[158],"outperform":[161],"current":[162],"state-of-the-art":[163],"substantially":[165],"lower":[166],"requirements.":[168],"On":[169],"CIFAR-10,":[170],"an":[173],"up":[176],"9%":[178],"accuracy":[180],"while":[181],"reducing":[182],"computation":[184],"time":[185],"than":[188],"85%.":[189],"Our":[190],"results":[191],"are":[192],"step":[194],"towards":[195],"architectures":[198],"make":[200],"optimal":[201],"use":[202],"their":[204],"bridge":[207],"privacy-utility":[209],"gap":[210],"between":[211],"non-private":[214],"computer":[218],"vision.":[219]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}