{"id":"https://openalex.org/W6948152654","doi":"https://doi.org/10.5075/epfl-thesis-9825","title":"Theory of Deep Learning: Neural Tangent Kernel and Beyond","display_name":"Theory of Deep Learning: Neural Tangent Kernel and Beyond","publication_year":2022,"publication_date":"2022-01-01","ids":{"openalex":"https://openalex.org/W6948152654","doi":"https://doi.org/10.5075/epfl-thesis-9825"},"language":"en","primary_location":{"id":"pmh:oai:infoscience.epfl.ch:295831","is_oa":true,"landing_page_url":"https://infoscience.epfl.ch/handle/20.500.14299/189941","pdf_url":null,"source":{"id":"https://openalex.org/S4306400487","display_name":"Infoscience (Ecole Polytechnique F\u00e9d\u00e9rale de Lausanne)","issn_l":null,"issn":null,"is_oa":true,"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":null,"raw_type":"doctoral thesis"},"type":"dissertation","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://infoscience.epfl.ch/handle/20.500.14299/189941","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Jacot-Guillarmod, Arthur Ulysse","orcid":null},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Jacot-Guillarmod, Arthur Ulysse","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":true,"primary_topic":{"id":"https://openalex.org/T10895","display_name":"Species Distribution and Climate Change","score":0.14249999821186066,"subfield":{"id":"https://openalex.org/subfields/2302","display_name":"Ecological Modeling"},"field":{"id":"https://openalex.org/fields/23","display_name":"Environmental Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10895","display_name":"Species Distribution and Climate Change","score":0.14249999821186066,"subfield":{"id":"https://openalex.org/subfields/2302","display_name":"Ecological Modeling"},"field":{"id":"https://openalex.org/fields/23","display_name":"Environmental Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T13568","display_name":"Wood and Agarwood Research","score":0.08630000054836273,"subfield":{"id":"https://openalex.org/subfields/1605","display_name":"Organic Chemistry"},"field":{"id":"https://openalex.org/fields/16","display_name":"Chemistry"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T13130","display_name":"Biological and pharmacological studies of plants","score":0.038100000470876694,"subfield":{"id":"https://openalex.org/subfields/2736","display_name":"Pharmacology"},"field":{"id":"https://openalex.org/fields/27","display_name":"Medicine"},"domain":{"id":"https://openalex.org/domains/4","display_name":"Health Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.6567999720573425},{"id":"https://openalex.org/keywords/hessian-matrix","display_name":"Hessian matrix","score":0.5697000026702881},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.5408999919891357},{"id":"https://openalex.org/keywords/tangent","display_name":"Tangent","score":0.5217999815940857},{"id":"https://openalex.org/keywords/kernel","display_name":"Kernel (algebra)","score":0.508400022983551},{"id":"https://openalex.org/keywords/convergence","display_name":"Convergence (economics)","score":0.49869999289512634},{"id":"https://openalex.org/keywords/simple","display_name":"Simple (philosophy)","score":0.4706999957561493},{"id":"https://openalex.org/keywords/function","display_name":"Function (biology)","score":0.44760000705718994},{"id":"https://openalex.org/keywords/kernel-method","display_name":"Kernel method","score":0.3930000066757202},{"id":"https://openalex.org/keywords/gradient-descent","display_name":"Gradient descent","score":0.36980000138282776}],"concepts":[{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.6567999720573425},{"id":"https://openalex.org/C203616005","wikidata":"https://www.wikidata.org/wiki/Q620495","display_name":"Hessian matrix","level":2,"score":0.5697000026702881},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.5408999919891357},{"id":"https://openalex.org/C138187205","wikidata":"https://www.wikidata.org/wiki/Q131251","display_name":"Tangent","level":2,"score":0.5217999815940857},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.515500009059906},{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.508400022983551},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.49869999289512634},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.4706999957561493},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.47040000557899475},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.44760000705718994},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.42149999737739563},{"id":"https://openalex.org/C122280245","wikidata":"https://www.wikidata.org/wiki/Q620622","display_name":"Kernel method","level":3,"score":0.3930000066757202},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.36980000138282776},{"id":"https://openalex.org/C158693339","wikidata":"https://www.wikidata.org/wiki/Q190524","display_name":"Eigenvalues and eigenvectors","level":2,"score":0.366100013256073},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3562999963760376},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.35359999537467957},{"id":"https://openalex.org/C2984842247","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep neural networks","level":3,"score":0.3504999876022339},{"id":"https://openalex.org/C106487976","wikidata":"https://www.wikidata.org/wiki/Q685816","display_name":"Matrix (chemical analysis)","level":2,"score":0.33719998598098755},{"id":"https://openalex.org/C188198153","wikidata":"https://www.wikidata.org/wiki/Q1613840","display_name":"Limiting","level":2,"score":0.3073999881744385},{"id":"https://openalex.org/C64812099","wikidata":"https://www.wikidata.org/wiki/Q176604","display_name":"Random matrix","level":3,"score":0.30480000376701355},{"id":"https://openalex.org/C2781238097","wikidata":"https://www.wikidata.org/wiki/Q175026","display_name":"Object (grammar)","level":2,"score":0.2662999927997589},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.26269999146461487},{"id":"https://openalex.org/C202444582","wikidata":"https://www.wikidata.org/wiki/Q837863","display_name":"Pure mathematics","level":1,"score":0.26089999079704285},{"id":"https://openalex.org/C2776799497","wikidata":"https://www.wikidata.org/wiki/Q484298","display_name":"Surface (topology)","level":2,"score":0.2606000006198883},{"id":"https://openalex.org/C207390915","wikidata":"https://www.wikidata.org/wiki/Q1230525","display_name":"Divergence (linguistics)","level":2,"score":0.26030001044273376},{"id":"https://openalex.org/C149629883","wikidata":"https://www.wikidata.org/wiki/Q660926","display_name":"Fraction (chemistry)","level":2,"score":0.25949999690055847},{"id":"https://openalex.org/C169756996","wikidata":"https://www.wikidata.org/wiki/Q194919","display_name":"Eigendecomposition of a matrix","level":3,"score":0.2572999894618988},{"id":"https://openalex.org/C151201525","wikidata":"https://www.wikidata.org/wiki/Q177239","display_name":"Limit (mathematics)","level":2,"score":0.2500999867916107},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.2484000027179718},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.24089999496936798},{"id":"https://openalex.org/C32277403","wikidata":"https://www.wikidata.org/wiki/Q740445","display_name":"Ridge","level":2,"score":0.2395000010728836},{"id":"https://openalex.org/C2777052490","wikidata":"https://www.wikidata.org/wiki/Q5072826","display_name":"Chaotic","level":2,"score":0.23819999396800995},{"id":"https://openalex.org/C156778621","wikidata":"https://www.wikidata.org/wiki/Q1365748","display_name":"Spectrum (functional analysis)","level":2,"score":0.23229999840259552},{"id":"https://openalex.org/C20538924","wikidata":"https://www.wikidata.org/wiki/Q822139","display_name":"Characteristic function (probability theory)","level":3,"score":0.23070000112056732},{"id":"https://openalex.org/C165263651","wikidata":"https://www.wikidata.org/wiki/Q2443460","display_name":"Topological conjugacy","level":2,"score":0.2282000035047531},{"id":"https://openalex.org/C202615002","wikidata":"https://www.wikidata.org/wiki/Q783507","display_name":"Differentiable function","level":2,"score":0.22599999606609344},{"id":"https://openalex.org/C184720557","wikidata":"https://www.wikidata.org/wiki/Q7825049","display_name":"Topology (electrical circuits)","level":2,"score":0.22470000386238098},{"id":"https://openalex.org/C118615104","wikidata":"https://www.wikidata.org/wiki/Q121416","display_name":"Discrete mathematics","level":1,"score":0.21979999542236328},{"id":"https://openalex.org/C57869625","wikidata":"https://www.wikidata.org/wiki/Q1783502","display_name":"Rate of convergence","level":3,"score":0.2134000062942505},{"id":"https://openalex.org/C191399111","wikidata":"https://www.wikidata.org/wiki/Q64861","display_name":"Transitive relation","level":2,"score":0.21209999918937683}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:infoscience.epfl.ch:295831","is_oa":true,"landing_page_url":"https://infoscience.epfl.ch/handle/20.500.14299/189941","pdf_url":null,"source":{"id":"https://openalex.org/S4306400487","display_name":"Infoscience (Ecole Polytechnique F\u00e9d\u00e9rale de Lausanne)","issn_l":null,"issn":null,"is_oa":true,"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":null,"raw_type":"doctoral thesis"},{"id":"doi:10.5075/epfl-thesis-9825","is_oa":true,"landing_page_url":"https://doi.org/10.5075/epfl-thesis-9825","pdf_url":null,"source":null,"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Dissertation"}],"best_oa_location":{"id":"pmh:oai:infoscience.epfl.ch:295831","is_oa":true,"landing_page_url":"https://infoscience.epfl.ch/handle/20.500.14299/189941","pdf_url":null,"source":{"id":"https://openalex.org/S4306400487","display_name":"Infoscience (Ecole Polytechnique F\u00e9d\u00e9rale de Lausanne)","issn_l":null,"issn":null,"is_oa":true,"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":null,"raw_type":"doctoral thesis"},"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":{"In":[0,31,78,237,289,343,384],"the":[1,57,60,67,70,74,79,84,89,92,97,114,124,138,142,146,149,158,166,172,175,189,220,223,226,251,256,259,275,278,300,311,344,348,360,364,367,376,387,391,403,406,431,454],"recent":[2],"years,":[3],"Deep":[4,422],"Neural":[5,61],"Networks":[6,296,424],"(DNNs)":[7],"have":[8],"managed":[9],"to":[10,65,95,100,107,122,132,281,306,332,337,363,386,446],"succeed":[11],"at":[12],"tasks":[13],"that":[14],"previously":[15],"appeared":[16],"impossible,":[17],"such":[18,292,418],"as":[19,293,369],"human-level":[20],"object":[21],"recognition,":[22],"text":[23],"synthesis,":[24],"translation,":[25],"playing":[26],"games":[27],"and":[28,103,136,196,244,255,261,287,313,352,372,380],"many":[29],"more.":[30],"spite":[32],"of":[33,39,44,59,69,86,91,113,116,128,130,141,145,148,152,169,174,211,217,222,225,234,253,272,274,277,310,359,366,405,415,437,442],"these":[34,40,264],"major":[35],"achievements,":[36],"our":[37],"understanding":[38],"models,":[41],"in":[42,88,164,206,324,341,412,421,453],"particular":[43],"what":[45],"happens":[46],"during":[47,76,395],"their":[48],"training,":[49,396],"remains":[50,393],"very":[51,238,427],"limited.":[52],"This":[53,119],"PhD":[54],"started":[55],"with":[56,425],"introduction":[58],"Tangent":[62],"Kernel":[63,197],"(NTK)":[64],"describe":[66],"evolution":[68,404],"function":[71,190],"represented":[72],"by":[73,192,231,250,269],"network":[75,93,177,301],"training.":[77,155],"infinite-width":[80,326],"limit,":[81,105],"i.e.":[82,171],"when":[83],"number":[85,233,414],"neurons":[87],"layers":[90],"grows":[94],"infinity,":[96],"NTK":[98,159,182,312,346,388,392,407,455],"converges":[99],"a":[101,108,133,161,185,232,245,282,357,413,426,435,447],"deterministic":[102],"time-independent":[104],"leading":[106,280,445],"simple":[109],"yet":[110],"complete":[111,321],"description":[112,140,322],"dynamics":[115,433],"infinitely-wide":[117],"DNNs.":[118],"allowed":[120],"one":[121],"give":[123,318],"first":[125,139],"general":[126],"proof":[127],"convergence":[129,285],"DNNs":[131,153,195,323,354,379],"global":[134],"minimum,":[135],"yielded":[137],"limiting":[143],"spectrum":[144],"Hessian":[147],"loss":[150],"surface":[151],"throughout":[154],"More":[156],"importantly,":[157],"plays":[160],"crucial":[162],"role":[163],"describing":[165],"generalization":[167,202],"abilities":[168],"DNNs,":[170],"performance":[173],"trained":[176],"on":[178,375],"unseen":[179],"data.":[180],"The":[181],"analysis":[183,216],"uncovered":[184],"direct":[186],"link":[187],"between":[188,258,284,350,378],"learned":[191],"infinitely":[193],"wide":[194],"Ridge":[198],"Regression":[199],"predictors,":[200],"whose":[201],"properties":[203,309],"are":[204,267],"studied":[205],"this":[207,325],"thesis":[208],"using":[209],"tools":[210],"random":[212,381],"matrix":[213],"theory.":[214],"Our":[215],"KRR":[218],"reveals":[219],"importance":[221],"eigendecomposition":[224],"NTK,":[227,279],"which":[228,410,450],"is":[229,329,355,408,451],"affected":[230],"architectural":[235],"choices.":[236],"deep":[239],"networks,":[240],"an":[241,319],"ordered":[242],"regime":[243,247,419],"chaotic":[246],"appear,":[248],"determined":[249],"choice":[252],"non-linearity":[254],"balance":[257],"weights":[260],"bias":[262,449],"parameters;":[263],"two":[265],"phases":[266],"characterized":[268],"different":[270],"speeds":[271],"decay":[273],"eigenvalues":[276],"tradeoff":[283],"speed":[286],"generalization.":[288],"practical":[290],"contexts":[291],"Generative":[294],"Adversarial":[295],"or":[297],"Topology":[298],"Optimization,":[299],"architecture":[302],"can":[303],"be":[304],"chosen":[305],"guarantee":[307],"certain":[308],"its":[314],"spectrum.":[315],"These":[316],"results":[317],"almost":[320],"limit.":[327],"It":[328],"then":[330],"natural":[331],"wonder":[333],"how":[334],"it":[335],"extends":[336],"finite-width":[338],"networks":[339],"used":[340],"practice.":[342],"so-called":[345,399],"regime,":[347,389],"discrepancy":[349],"finite-":[351],"infinite-widths":[353],"mainly":[356],"result":[358],"variance":[361],"w.r.t.":[362],"sampling":[365],"parameters,":[368],"shown":[370],"empirically":[371],"mathematically":[373],"relying":[374],"similarity":[377],"feature":[382],"models.":[383],"contrast":[385],"where":[390,402,430],"constant":[394],"there":[397],"exist":[398],"active":[400],"regimes,":[401],"significant,":[409],"appear":[411],"settings.":[416],"One":[417],"appears":[420],"Linear":[423],"small":[428],"initialization,":[429],"training":[432],"approach":[434],"sequence":[436],"saddle-points,":[438],"representing":[439],"linear":[440],"maps":[441],"increasing":[443],"rank,":[444],"low-rank":[448],"absent":[452],"regime.":[456]},"counts_by_year":[],"updated_date":"2026-07-23T05:56:39.545243","created_date":"2025-10-10T00:00:00"}