{"id":"https://openalex.org/W7214786452","doi":"https://doi.org/10.48550/arxiv.2609.35377","title":"First Learn, Then Memorize: The Spectral Bias of Diffusion Models","display_name":"First Learn, Then Memorize: The Spectral Bias of Diffusion Models","publication_year":2026,"publication_date":"2026-09-28","ids":{"openalex":"https://openalex.org/W7214786452","doi":"https://doi.org/10.48550/arxiv.2609.35377"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.35377","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.35377","pdf_url":"https://arxiv.org/pdf/2609.35377","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":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.35377","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5145235847","display_name":"Rapha\u00ebl Urfin","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I1294671590","display_name":"Centre National de la Recherche Scientifique","ror":"https://ror.org/02feahw73","country_code":"FR","type":"government","lineage":["https://openalex.org/I1294671590"]},{"id":"https://openalex.org/I204730241","display_name":"Universit\u00e9 Paris Cit\u00e9","ror":"https://ror.org/05f82e368","country_code":"FR","type":"education","lineage":["https://openalex.org/I204730241"]},{"id":"https://openalex.org/I2746051580","display_name":"Universit\u00e9 Paris Sciences et Lettres","ror":"https://ror.org/013cjyk83","country_code":"FR","type":"education","lineage":["https://openalex.org/I2746051580"]},{"id":"https://openalex.org/I29607241","display_name":"\u00c9cole Normale Sup\u00e9rieure - PSL","ror":"https://ror.org/05a0dhs15","country_code":"FR","type":"other","lineage":["https://openalex.org/I2746051580","https://openalex.org/I29607241"]},{"id":"https://openalex.org/I39804081","display_name":"Sorbonne Universit\u00e9","ror":"https://ror.org/02en5vm52","country_code":"FR","type":"education","lineage":["https://openalex.org/I39804081"]},{"id":"https://openalex.org/I4210129651","display_name":"Laboratoire de Physique de l'Ecole Normale Sup\u00e9rieure","ror":"https://ror.org/03a26mh11","country_code":"FR","type":"facility","lineage":["https://openalex.org/I1294671590","https://openalex.org/I204730241","https://openalex.org/I2746051580","https://openalex.org/I2800004676","https://openalex.org/I29607241","https://openalex.org/I39804081","https://openalex.org/I4210129651"]}],"countries":["FR"],"is_corresponding":false,"raw_author_name":"Urfin, Rapha\u00ebl","raw_affiliation_strings":["Laboratoire de Physique de l'\u00c9cole normale sup\u00e9rieure , ENS , Universit\u00e9 PSL , CNRS , Sorbonne Universit\u00e9 , Universit\u00e9 Paris Cit\u00e9 , F-75005 Paris , France"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Laboratoire de Physique de l'\u00c9cole normale sup\u00e9rieure , ENS , Universit\u00e9 PSL , CNRS , Sorbonne Universit\u00e9 , Universit\u00e9 Paris Cit\u00e9 , F-75005 Paris , France","institution_ids":["https://openalex.org/I1294671590","https://openalex.org/I204730241","https://openalex.org/I2746051580","https://openalex.org/I29607241","https://openalex.org/I39804081","https://openalex.org/I4210129651"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5029435863","display_name":"Tony Bonnaire","orcid":"https://orcid.org/0000-0003-2149-8795","observed_orcids":["https://orcid.org/0000-0003-2149-8795"]},"institutions":[{"id":"https://openalex.org/I1294671590","display_name":"Centre National de la Recherche Scientifique","ror":"https://ror.org/02feahw73","country_code":"FR","type":"government","lineage":["https://openalex.org/I1294671590"]},{"id":"https://openalex.org/I277688954","display_name":"Universit\u00e9 Paris-Saclay","ror":"https://ror.org/03xjwb503","country_code":"FR","type":"education","lineage":["https://openalex.org/I277688954"]},{"id":"https://openalex.org/I4210103272","display_name":"Institut d'Astrophysique Spatiale","ror":"https://ror.org/014p8mr66","country_code":"FR","type":"facility","lineage":["https://openalex.org/I1294671590","https://openalex.org/I277688954","https://openalex.org/I4210103272"]}],"countries":["FR"],"is_corresponding":false,"raw_author_name":"Bonnaire, Tony","raw_affiliation_strings":["Universit\u00e9 Paris-Saclay , CNRS , Institut d'Astrophysique Spatiale , 91405 Orsay , France"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Universit\u00e9 Paris-Saclay , CNRS , Institut d'Astrophysique Spatiale , 91405 Orsay , France","institution_ids":["https://openalex.org/I1294671590","https://openalex.org/I277688954","https://openalex.org/I4210103272"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5002294199","display_name":"Giulio Biroli","orcid":"https://orcid.org/0000-0003-4115-2644","observed_orcids":["https://orcid.org/0000-0003-4115-2644"]},"institutions":[{"id":"https://openalex.org/I1294671590","display_name":"Centre National de la Recherche Scientifique","ror":"https://ror.org/02feahw73","country_code":"FR","type":"government","lineage":["https://openalex.org/I1294671590"]},{"id":"https://openalex.org/I204730241","display_name":"Universit\u00e9 Paris Cit\u00e9","ror":"https://ror.org/05f82e368","country_code":"FR","type":"education","lineage":["https://openalex.org/I204730241"]},{"id":"https://openalex.org/I2746051580","display_name":"Universit\u00e9 Paris Sciences et Lettres","ror":"https://ror.org/013cjyk83","country_code":"FR","type":"education","lineage":["https://openalex.org/I2746051580"]},{"id":"https://openalex.org/I29607241","display_name":"\u00c9cole Normale Sup\u00e9rieure - PSL","ror":"https://ror.org/05a0dhs15","country_code":"FR","type":"other","lineage":["https://openalex.org/I2746051580","https://openalex.org/I29607241"]},{"id":"https://openalex.org/I39804081","display_name":"Sorbonne Universit\u00e9","ror":"https://ror.org/02en5vm52","country_code":"FR","type":"education","lineage":["https://openalex.org/I39804081"]},{"id":"https://openalex.org/I4210129651","display_name":"Laboratoire de Physique de l'Ecole Normale Sup\u00e9rieure","ror":"https://ror.org/03a26mh11","country_code":"FR","type":"facility","lineage":["https://openalex.org/I1294671590","https://openalex.org/I204730241","https://openalex.org/I2746051580","https://openalex.org/I2800004676","https://openalex.org/I29607241","https://openalex.org/I39804081","https://openalex.org/I4210129651"]}],"countries":["FR"],"is_corresponding":false,"raw_author_name":"Biroli, Giulio","raw_affiliation_strings":["Laboratoire de Physique de l'\u00c9cole normale sup\u00e9rieure , ENS , Universit\u00e9 PSL , CNRS , Sorbonne Universit\u00e9 , Universit\u00e9 Paris Cit\u00e9 , F-75005 Paris , France"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Laboratoire de Physique de l'\u00c9cole normale sup\u00e9rieure , ENS , Universit\u00e9 PSL , CNRS , Sorbonne Universit\u00e9 , Universit\u00e9 Paris Cit\u00e9 , F-75005 Paris , France","institution_ids":["https://openalex.org/I1294671590","https://openalex.org/I204730241","https://openalex.org/I2746051580","https://openalex.org/I29607241","https://openalex.org/I39804081","https://openalex.org/I4210129651"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5037000437","display_name":"Marc M\u00e9zard","orcid":"https://orcid.org/0000-0003-4591-3893","observed_orcids":["https://orcid.org/0000-0003-4591-3893"]},"institutions":[{"id":"https://openalex.org/I71209653","display_name":"Bocconi University","ror":"https://ror.org/05crjpb27","country_code":"IT","type":"education","lineage":["https://openalex.org/I71209653"]}],"countries":["IT"],"is_corresponding":false,"raw_author_name":"M\u00e9zard, Marc","raw_affiliation_strings":["Department of Computing Sciences , Bocconi University , Milano , Italy"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department of Computing Sciences , Bocconi University , Milano , Italy","institution_ids":["https://openalex.org/I71209653"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":9,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.4041000008583069,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.4041000008583069,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.2865999937057495,"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/T10320","display_name":"Neural Networks and Applications","score":0.04309999942779541,"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":[{"id":"https://openalex.org/keywords/diffusion-models","display_name":"diffusion models","score":0.9440000057220459},{"id":"https://openalex.org/keywords/spectral-bias","display_name":"spectral bias","score":0.8149999976158142},{"id":"https://openalex.org/keywords/memorization","display_name":"memorization","score":0.7910000085830688},{"id":"https://openalex.org/keywords/generalization","display_name":"generalization","score":0.9419999718666077},{"id":"https://openalex.org/keywords/neural-tangent-kernel","display_name":"neural tangent kernel","score":0.9079999923706055},{"id":"https://openalex.org/keywords/score-matching","display_name":"score matching","score":0.8600000143051147},{"id":"https://openalex.org/keywords/bias-variance-decomposition","display_name":"bias-variance decomposition","score":0.9079999923706055},{"id":"https://openalex.org/keywords/high-dimensional-limit","display_name":"high-dimensional limit","score":0.47200000286102295},{"id":"https://openalex.org/keywords/u-net","display_name":"U-Net","score":0.7609999775886536},{"id":"https://openalex.org/keywords/training-set-size","display_name":"training data size","score":0.35100001096725464}],"concepts":[{"id":"https://openalex.org/C74193536","wikidata":"https://www.wikidata.org/wiki/Q574844","display_name":"Kernel (algebra)","level":2,"score":0.6262999773025513},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.6176000237464905},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.6118000149726868},{"id":"https://openalex.org/C158693339","wikidata":"https://www.wikidata.org/wiki/Q190524","display_name":"Eigenvalues and eigenvectors","level":2,"score":0.5878999829292297},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5321999788284302},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.5080000162124634},{"id":"https://openalex.org/C90119067","wikidata":"https://www.wikidata.org/wiki/Q43260","display_name":"Polynomial","level":2,"score":0.490200012922287},{"id":"https://openalex.org/C106487976","wikidata":"https://www.wikidata.org/wiki/Q685816","display_name":"Matrix (chemical analysis)","level":2,"score":0.4422999918460846},{"id":"https://openalex.org/C151201525","wikidata":"https://www.wikidata.org/wiki/Q177239","display_name":"Limit (mathematics)","level":2,"score":0.40939998626708984},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.3939000070095062},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.3928000032901764},{"id":"https://openalex.org/C156778621","wikidata":"https://www.wikidata.org/wiki/Q1365748","display_name":"Spectrum (functional analysis)","level":2,"score":0.38999998569488525},{"id":"https://openalex.org/C198531522","wikidata":"https://www.wikidata.org/wiki/Q485146","display_name":"Sample (material)","level":2,"score":0.38370001316070557},{"id":"https://openalex.org/C2776359362","wikidata":"https://www.wikidata.org/wiki/Q2145286","display_name":"Representation (politics)","level":3,"score":0.3589000105857849},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.35850000381469727},{"id":"https://openalex.org/C122280245","wikidata":"https://www.wikidata.org/wiki/Q620622","display_name":"Kernel method","level":3,"score":0.3456999957561493},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.34389999508857727},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.33880001306533813},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.3287999927997589},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.32600000500679016},{"id":"https://openalex.org/C110121322","wikidata":"https://www.wikidata.org/wiki/Q865811","display_name":"Distribution (mathematics)","level":2,"score":0.32199999690055847},{"id":"https://openalex.org/C45347329","wikidata":"https://www.wikidata.org/wiki/Q5166604","display_name":"Convolution (computer science)","level":3,"score":0.31060001254081726},{"id":"https://openalex.org/C77246614","wikidata":"https://www.wikidata.org/wiki/Q1409400","display_name":"Gramian matrix","level":3,"score":0.30959999561309814},{"id":"https://openalex.org/C162392398","wikidata":"https://www.wikidata.org/wiki/Q272404","display_name":"Finite set","level":2,"score":0.30880001187324524},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.30880001187324524},{"id":"https://openalex.org/C135692309","wikidata":"https://www.wikidata.org/wiki/Q111124","display_name":"Square (algebra)","level":2,"score":0.2694999873638153}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.35377","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.35377","pdf_url":"https://arxiv.org/pdf/2609.35377","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":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2609.35377","doi":"https://doi.org/10.48550/arxiv.2609.35377","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.35377","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2609.35377","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.35377","pdf_url":"https://arxiv.org/pdf/2609.35377","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":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G2439744834","display_name":"PR[AI]RIE-PSAI - Paris School of Artificial Intelligence","funder_award_id":"ANR-23-IACL-0008","funder_id":"https://openalex.org/F4320320883","funder_display_name":"Agence Nationale de la Recherche"},{"id":"https://openalex.org/G534385645","display_name":"Mathematics of Generation: an Interdisciplinary Cross Analysis of Loss Landscapes","funder_award_id":"ANR-25-PEIA-0004","funder_id":"https://openalex.org/F4320320883","funder_display_name":"Agence Nationale de la Recherche"},{"id":"https://openalex.org/G7080048889","display_name":null,"funder_award_id":"ANR-23","funder_id":"https://openalex.org/F4320320883","funder_display_name":"Agence Nationale de la Recherche"}],"funders":[{"id":"https://openalex.org/F4320320883","display_name":"Agence Nationale de la Recherche","ror":"https://ror.org/00rbzpz17"},{"id":"https://openalex.org/F4320326256","display_name":"Grand \u00c9quipement National De Calcul Intensif","ror":"https://ror.org/0010d1q40"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7214786452.pdf","grobid_xml":"https://content.openalex.org/works/W7214786452.grobid-xml"},"referenced_works_count":33,"referenced_works":["https://openalex.org/W1834627138","https://openalex.org/W1991111872","https://openalex.org/W2013035813","https://openalex.org/W2014747354","https://openalex.org/W2039953809","https://openalex.org/W2090816344","https://openalex.org/W2117756735","https://openalex.org/W2160733696","https://openalex.org/W2590106345","https://openalex.org/W2952204734","https://openalex.org/W2988294134","https://openalex.org/W3172995164","https://openalex.org/W4224815527","https://openalex.org/W4287752094","https://openalex.org/W4297676498","https://openalex.org/W4312426305","https://openalex.org/W4390832399","https://openalex.org/W4394881816","https://openalex.org/W4403444482","https://openalex.org/W4405031167","https://openalex.org/W4406301905","https://openalex.org/W4408869991","https://openalex.org/W4412033860","https://openalex.org/W4413182660","https://openalex.org/W4415368639","https://openalex.org/W4416150179","https://openalex.org/W4416374356","https://openalex.org/W6966946868","https://openalex.org/W6979330738","https://openalex.org/W7133208375","https://openalex.org/W7133923690","https://openalex.org/W7164848538","https://openalex.org/W7204250564"],"related_works":["https://openalex.org/W7203757531","https://openalex.org/W6948114292","https://openalex.org/W4416603978","https://openalex.org/W4412033860","https://openalex.org/W4407123503","https://openalex.org/W7125938874","https://openalex.org/W7077894800","https://openalex.org/W7204250564","https://openalex.org/W4406884100","https://openalex.org/W4416453961"],"abstract_inverted_index":{"Diffusion":[0],"models":[1],"trained":[2,252],"on":[3,61,162,187,246],"a":[4,109,172,216],"finite":[5],"dataset":[6],"first":[7,221],"learn":[8],"to":[9],"generate":[10],"novel,":[11],"high-quality":[12],"samples":[13],"and":[14,32,47,71,85,141,159,206,213,248,258,275],"only":[15],"much":[16],"later":[17],"collapse":[18],"onto":[19],"their":[20],"training":[21,39,64,179],"set.":[22],"We":[23,80,183],"identify":[24],"the":[25,33,42,51,55,62,67,87,114,120,134,138,150,156,166,178,193,196,220,224,228,231,239,255,261,265,272,280],"mechanism":[26],"behind":[27],"this":[28,185],"separation":[29],"of":[30,41,54,69,72,99,130,137,155],"timescales":[31,68],"object":[34],"that":[35,82,86,219],"probes":[36],"it.":[37],"The":[38,97,128,147],"dynamics":[40],"score":[43],"function":[44],"are":[45],"governed---exactly,":[46],"at":[48,108,268],"any":[49],"width---by":[50],"Gram":[52,121,266],"matrix":[53,122,267],"Neural":[56],"Tangent":[57],"Kernel":[58],"(NTK)":[59],"evaluated":[60],"noisy":[63],"data,":[65],"so":[66],"generalization":[70],"memorization":[73,173,284],"must":[74],"be":[75],"encoded":[76],"in":[77,93,113,177,195,243,249,285],"its":[78],"spectrum.":[79],"show":[81,238],"they":[83],"are,":[84],"structure":[88,242],"responsible":[89],"has":[90],"no":[91],"analogue":[92],"standard":[94],"kernel":[95],"settings.":[96],"use":[98],"multiple":[100],"noise":[101,111,168],"realizations":[102],"per":[103],"sample":[104,211],"($m$":[105],"noised":[106],"copies":[107],"fixed":[110],"level)":[112],"score-matching":[115],"loss":[116],"is":[117,142,160],"what":[118],"restructures":[119],"spectrum":[123,194],"into":[124],"two":[125,188],"distinct":[126],"parts.":[127],"first,":[129],"large":[131],"eigenvalues,":[132],"carries":[133],"global":[135],"features":[136],"target":[139],"distribution":[140],"present":[143],"already":[144],"for":[145,200],"$m=1$.":[146],"second,":[148],"which":[149],"repeated":[151],"noising":[152],"creates,":[153],"consists":[154],"smallest":[157],"eigenvalues":[158],"supported":[161],"eigenvectors":[163],"aligned":[164],"with":[165,234],"sample-specific":[167],"directions;":[169],"it":[170],"sets":[171],"timescale":[174],"parametrically":[175],"larger":[176],"set":[180],"size":[181],"$n$.":[182],"establish":[184],"picture":[186],"fronts.":[189],"Analytically,":[190],"we":[191,237,259],"solve":[192],"lazy":[197,256],"high-dimensional":[198],"limit":[199],"both":[201],"linear":[202],"($n":[203,208],"\\asymp":[204,209],"d$)":[205],"polynomial":[207],"d^k$)":[210],"complexities,":[212],"prove":[214],"through":[215],"bias--variance":[217],"decomposition":[218],"bulk":[222,282],"minimizes":[223],"approximation":[225],"error":[226,232],"while":[227],"second":[229,281],"drives":[230],"associated":[233],"memorization.":[235],"Empirically,":[236],"same":[240],"two-bulk":[241],"Convolutional":[244],"NTKs":[245],"CelebA":[247],"finite-width":[250],"U-Nets":[251],"well":[253],"beyond":[254],"regime,":[257],"make":[260],"link":[262],"causal:":[263],"truncating":[264],"rank":[269],"$r$":[270],"tunes":[271],"generalization--memorization":[273],"transition,":[274],"an":[276],"$L_2$":[277],"penalty":[278],"targeting":[279],"suppresses":[283],"feature-learning":[286],"U-Nets.":[287]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-09-30T00:00:00"}