{"id":"https://openalex.org/W4392866992","doi":"https://doi.org/10.48550/arxiv.2403.08938","title":"A non-asymptotic theory of Kernel Ridge Regression: deterministic equivalents, test error, and GCV estimator","display_name":"A non-asymptotic theory of Kernel Ridge Regression: deterministic equivalents, test error, and GCV estimator","publication_year":2024,"publication_date":"2024-03-13","ids":{"openalex":"https://openalex.org/W4392866992","doi":"https://doi.org/10.48550/arxiv.2403.08938"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2403.08938","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2403.08938","pdf_url":"https://arxiv.org/pdf/2403.08938","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2403.08938","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5063392474","display_name":"Theodor Misiakiewicz","orcid":"https://orcid.org/0000-0003-1470-7536","observed_orcids":["https://orcid.org/0000-0003-1470-7536"]},"institutions":[{"id":"https://openalex.org/I160992636","display_name":"Toyota Technological Institute at Chicago","ror":"https://ror.org/02sn5gb64","country_code":"US","type":"education","lineage":["https://openalex.org/I160992636"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Misiakiewicz, Theodor","raw_affiliation_strings":["Toyota Technological Institute at Chicago"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Toyota Technological Institute at Chicago","institution_ids":["https://openalex.org/I160992636"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5047351142","display_name":"Basil Saeed","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I97018004","display_name":"Stanford University","ror":"https://ror.org/00f54p054","country_code":"US","type":"education","lineage":["https://openalex.org/I97018004"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Saeed, Basil","raw_affiliation_strings":["Department of Electrical Engineering , Stanford University"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Electrical Engineering , Stanford University","institution_ids":["https://openalex.org/I97018004"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":1,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.17520000040531158,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.17520000040531158,"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/T10136","display_name":"Statistical Methods and Inference","score":0.1737000048160553,"subfield":{"id":"https://openalex.org/subfields/2613","display_name":"Statistics and Probability"},"field":{"id":"https://openalex.org/fields/26","display_name":"Mathematics"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11871","display_name":"Advanced Statistical Methods and Models","score":0.12080000340938568,"subfield":{"id":"https://openalex.org/subfields/2613","display_name":"Statistics and Probability"},"field":{"id":"https://openalex.org/fields/26","display_name":"Mathematics"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/kernel-ridge-regression","display_name":"kernel ridge regression","score":1.0},{"id":"https://openalex.org/keywords/deterministic-equivalent","display_name":"deterministic equivalent","score":0.8569999933242798},{"id":"https://openalex.org/keywords/test-error","display_name":"test error","score":0.5820000171661377},{"id":"https://openalex.org/keywords/generalized-cross-validation","display_name":"generalized cross-validation","score":0.9169999957084656},{"id":"https://openalex.org/keywords/nonasymptotic-bounds","display_name":"non-asymptotic bounds","score":0.8489999771118164},{"id":"https://openalex.org/keywords/random-matrix-theory","display_name":"random matrix theory","score":0.7730000019073486},{"id":"https://openalex.org/keywords/regularization-parameter-selection","display_name":"regularization parameter selection","score":0.6240000128746033}],"concepts":[{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.7084832787513733},{"id":"https://openalex.org/C32277403","wikidata":"https://www.wikidata.org/wiki/Q740445","display_name":"Ridge","level":2,"score":0.6341345310211182},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.5585950613021851},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.5574411153793335},{"id":"https://openalex.org/C200695384","wikidata":"https://www.wikidata.org/wiki/Q1739319","display_name":"Kernel regression","level":3,"score":0.5212236642837524},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.4940105080604553},{"id":"https://openalex.org/C83546350","wikidata":"https://www.wikidata.org/wiki/Q1139051","display_name":"Regression","level":2,"score":0.45783117413520813},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.4170433282852173},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.385911762714386},{"id":"https://openalex.org/C127313418","wikidata":"https://www.wikidata.org/wiki/Q1069","display_name":"Geology","level":0,"score":0.20016714930534363},{"id":"https://openalex.org/C114614502","wikidata":"https://www.wikidata.org/wiki/Q76592","display_name":"Combinatorics","level":1,"score":0.07160776853561401},{"id":"https://openalex.org/C151730666","wikidata":"https://www.wikidata.org/wiki/Q7205","display_name":"Paleontology","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2403.08938","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2403.08938","pdf_url":"https://arxiv.org/pdf/2403.08938","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2403.08938","doi":"https://doi.org/10.48550/arxiv.2403.08938","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2403.08938","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":"pmh:oai:arXiv.org:2403.08938","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2403.08938","pdf_url":"https://arxiv.org/pdf/2403.08938","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4392866992.pdf","grobid_xml":"https://content.openalex.org/works/W4392866992.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4406301905","https://openalex.org/W7116963477","https://openalex.org/W3033355066","https://openalex.org/W6885133903","https://openalex.org/W4286907835","https://openalex.org/W3035950049","https://openalex.org/W4238222158","https://openalex.org/W3121358889","https://openalex.org/W4200337531","https://openalex.org/W4391591572"],"abstract_inverted_index":{"We":[0,194],"consider":[1],"learning":[2],"an":[3,59,204],"unknown":[4],"target":[5,160],"function":[6,161],"$f_*$":[7],"using":[8],"kernel":[9,71,128,224],"ridge":[10,249],"regression":[11],"(KRR)":[12],"given":[13],"i.i.d.":[14],"data":[15,270],"$(u_i,y_i)$,":[16],"$i\\leq":[17],"n$,":[18],"where":[19],"$u_i":[20],"\\in":[21,33],"U$":[22],"is":[23],"a":[24,54,74,115,136,173,246,259],"covariate":[25],"vector":[26],"and":[27,123,158,191,202,210,274],"$y_i":[28],"=":[29],"f_*":[30],"(u_i)":[31],"+\\varepsilon_i":[32],"\\mathbb{R}$.":[34],"A":[35],"recent":[36],"string":[37],"of":[38,47,69,118,144,166,176,222,248],"work":[39],"has":[40,80],"empirically":[41],"shown":[42],"that":[43,63,110,152,252],"the":[44,67,70,127,141,156,159,164,167,184,220,223,234,241,261,271],"test":[45,142,242,272],"error":[46,143,243,273],"KRR":[48,145],"can":[49,264],"be":[50,265],"well":[51],"approximated":[52],"by":[53,189],"closed-form":[55],"estimate":[56,268],"derived":[57],"from":[58,269],"`equivalent'":[60],"sequence":[61],"model":[62],"only":[64,153],"depends":[65,154],"on":[66,85,126,155,172,216,240],"spectrum":[68],"operator.":[72,225],"However,":[73],"theoretical":[75,208],"justification":[76],"for":[77,99,114,140,179,278],"this":[78,106,111,134,196,231],"equivalence":[79,112],"so":[81],"far":[82],"relied":[83],"either":[84],"restrictive":[86],"assumptions":[87],"--":[88,94,146,151],"such":[89],"as":[90],"subgaussian":[91],"independent":[92],"eigenfunctions":[93],",":[95],"or":[96],"asymptotic":[97],"derivations":[98],"specific":[100],"kernels":[101],"in":[102,133,183],"high":[103],"dimensions.":[104],"In":[105],"paper,":[107],"we":[108,131,227],"prove":[109,228],"holds":[113],"general":[116],"class":[117],"problems":[119],"satisfying":[120],"some":[121],"spectral":[122],"concentration":[124],"properties":[125],"eigendecomposition.":[129],"Specifically,":[130],"establish":[132],"setting":[135,197],"non-asymptotic":[137,149],"deterministic":[138,177],"approximation":[139],"with":[147],"explicit":[148],"bounds":[150],"eigenvalues":[157],"alignment":[162],"to":[163,198,219,267],"eigenvectors":[165],"kernel.":[168],"Our":[169],"proofs":[170],"rely":[171,215],"careful":[174],"derivation":[175],"equivalents":[178],"random":[180],"matrix":[181],"functionals":[182],"dimension":[185],"free":[186],"regime":[187],"pioneered":[188],"Cheng":[190],"Montanari":[192],"(2022).":[193],"apply":[195],"several":[199],"classical":[200],"examples":[201],"show":[203],"excellent":[205],"agreement":[206],"between":[207],"predictions":[209],"numerical":[211],"simulations.":[212],"These":[213],"results":[214],"having":[217],"access":[218],"eigendecomposition":[221],"Alternatively,":[226],"that,":[229],"under":[230],"same":[232],"setting,":[233],"generalized":[235],"cross-validation":[236],"(GCV)":[237],"estimator":[238,263],"concentrates":[239],"uniformly":[244],"over":[245],"range":[247],"regularization":[250,276],"parameter":[251,277],"includes":[253],"zero":[254],"(the":[255],"interpolating":[256],"solution).":[257],"As":[258],"consequence,":[260],"GCV":[262],"used":[266],"optimal":[275],"KRR.":[279]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2026-09-30T09:27:27.532063","created_date":"2024-03-16T00:00:00"}