{"id":"https://openalex.org/W7163003182","doi":"https://doi.org/10.48550/arxiv.2605.30642","title":"Diffusion Models Preferentially Memorize Prototypical Examples or: Why Does My Diffusion Model Love Slop?","display_name":"Diffusion Models Preferentially Memorize Prototypical Examples or: Why Does My Diffusion Model Love Slop?","publication_year":2026,"publication_date":"2026-05-28","ids":{"openalex":"https://openalex.org/W7163003182","doi":"https://doi.org/10.48550/arxiv.2605.30642"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2605.30642","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2605.30642","pdf_url":"https://arxiv.org/pdf/2605.30642","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":"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/2605.30642","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137533570","display_name":"Marta Aparicio Rodriguez","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rodriguez, Marta Aparicio","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5033819196","display_name":"Anastasia Borovykh","orcid":"https://orcid.org/0000-0002-8550-4827","observed_orcids":["https://orcid.org/0000-0002-8550-4827"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Borovykh, Anastasia","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5044857425","display_name":"Grigorios A. Pavliotis","orcid":"https://orcid.org/0000-0002-3468-9227","observed_orcids":["https://orcid.org/0000-0002-3468-9227"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Pavliotis, Grigorios A.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5033093356","display_name":"Daniel J. Korchinski","orcid":"https://orcid.org/0000-0002-4224-4607","observed_orcids":["https://orcid.org/0000-0002-4224-4607"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Korchinski, Daniel J.","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":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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.4219000041484833,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.4219000041484833,"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/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.23800000548362732,"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/T10764","display_name":"Privacy-Preserving Technologies in Data","score":0.13619999587535858,"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":0.8379999995231628},{"id":"https://openalex.org/keywords/memorization","display_name":"memorization","score":0.7509999871253967},{"id":"https://openalex.org/keywords/data-memorization","display_name":"data memorization","score":0.5609999895095825},{"id":"https://openalex.org/keywords/production-rules","display_name":"production rules","score":0.5910000205039978},{"id":"https://openalex.org/keywords/heavy-tailed-distributions","display_name":"heavy-tailed distributions","score":0.8510000109672546},{"id":"https://openalex.org/keywords/dataset-diversity","display_name":"data diversity","score":0.9079999923706055},{"id":"https://openalex.org/keywords/data-deduplication","display_name":"data deduplication","score":0.7749999761581421},{"id":"https://openalex.org/keywords/privacy","display_name":"privacy","score":0.6930000185966492},{"id":"https://openalex.org/keywords/mean-reversion","display_name":"mean reversion","score":0.7110000252723694}],"concepts":[{"id":"https://openalex.org/C30038468","wikidata":"https://www.wikidata.org/wiki/Q4354775","display_name":"Memorization","level":2,"score":0.8465999960899353},{"id":"https://openalex.org/C182407805","wikidata":"https://www.wikidata.org/wiki/Q2626534","display_name":"Substring","level":3,"score":0.8126999735832214},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6129000186920166},{"id":"https://openalex.org/C31170391","wikidata":"https://www.wikidata.org/wiki/Q188619","display_name":"Hierarchy","level":2,"score":0.6043000221252441},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5364999771118164},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.5351999998092651},{"id":"https://openalex.org/C2778348673","wikidata":"https://www.wikidata.org/wiki/Q739302","display_name":"Production (economics)","level":2,"score":0.4641000032424927},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.41690000891685486},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.40619999170303345},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.31949999928474426},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.31700000166893005},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3109999895095825},{"id":"https://openalex.org/C2780598303","wikidata":"https://www.wikidata.org/wiki/Q65921492","display_name":"Flexibility (engineering)","level":2,"score":0.30640000104904175},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.3059000074863434},{"id":"https://openalex.org/C160920958","wikidata":"https://www.wikidata.org/wiki/Q7662746","display_name":"Synthetic data","level":2,"score":0.30309998989105225},{"id":"https://openalex.org/C88871306","wikidata":"https://www.wikidata.org/wiki/Q7208287","display_name":"Point process","level":2,"score":0.29589998722076416},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.2924000024795532},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.29109999537467957},{"id":"https://openalex.org/C2777601683","wikidata":"https://www.wikidata.org/wiki/Q6499736","display_name":"Vocabulary","level":2,"score":0.27869999408721924},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2766000032424927},{"id":"https://openalex.org/C121864883","wikidata":"https://www.wikidata.org/wiki/Q677916","display_name":"Statistical physics","level":1,"score":0.27090001106262207},{"id":"https://openalex.org/C101104100","wikidata":"https://www.wikidata.org/wiki/Q1063540","display_name":"Heteroscedasticity","level":2,"score":0.2702000141143799},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.25429999828338623}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2605.30642","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2605.30642","pdf_url":"https://arxiv.org/pdf/2605.30642","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2605.30642","doi":"https://doi.org/10.48550/arxiv.2605.30642","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.30642","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:2605.30642","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2605.30642","pdf_url":"https://arxiv.org/pdf/2605.30642","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320320283","display_name":"Imperial College London","ror":"https://ror.org/041kmwe10"},{"id":"https://openalex.org/F4320334593","display_name":"Natural Sciences and Engineering Research Council of Canada","ror":"https://ror.org/01h531d29"}],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7163003182.pdf","grobid_xml":"https://content.openalex.org/works/W7163003182.grobid-xml"},"referenced_works_count":13,"referenced_works":["https://openalex.org/W3133702157","https://openalex.org/W3177765786","https://openalex.org/W4224035735","https://openalex.org/W4226057201","https://openalex.org/W4312933868","https://openalex.org/W4385801484","https://openalex.org/W4400195451","https://openalex.org/W4401201665","https://openalex.org/W4406884100","https://openalex.org/W4415145279","https://openalex.org/W4416453961","https://openalex.org/W7077065852","https://openalex.org/W7162651035"],"related_works":["https://openalex.org/W4415338189","https://openalex.org/W7222324016","https://openalex.org/W4387389994","https://openalex.org/W7125165027","https://openalex.org/W6910402483","https://openalex.org/W7221082883","https://openalex.org/W7077894800","https://openalex.org/W7164848538","https://openalex.org/W4417152753","https://openalex.org/W7131122214"],"abstract_inverted_index":{"Generative":[0],"models":[1,58,188],"have":[2],"a":[3,108],"persistent":[4],"limitation:":[5],"their":[6],"tendency":[7],"to":[8,63],"memorize":[9],"training":[10,90,182],"data":[11,91,102],"can":[12],"create":[13],"legal":[14],"liabilities":[15],"and":[16,30,73,176],"erode":[17],"creative":[18],"diversity.":[19],"Understanding":[20],"which":[21,170],"samples":[22,49,76],"are":[23,81,134,173],"memorized":[24,50],"in":[25,28,52,156,169,185],"whole":[26],"or":[27,47],"part,":[29],"under":[31],"what":[32],"conditions,":[33],"therefore":[34],"remains":[35],"an":[36,153,163],"important":[37,154],"open":[38],"problem.":[39],"Here":[40],"we":[41,113,161],"answer":[42],"the":[43,53,64,68,89,101,191],"question":[44],"\"Are":[45],"atypical":[46,126],"rare":[48],"first?\"":[51],"negative.":[54],"We":[55],"train":[56],"diffusion":[57],"on":[59],"strings":[60],"generated":[61],"according":[62],"production":[65,138],"rules":[66],"of":[67,78,93,150,166],"Random":[69],"Hierarchy":[70],"Model":[71],"(RHM),":[72],"find":[74],"that":[75,98,143],"composed":[77],"common":[79,171],"substrings":[80,172],"preferentially":[82],"memorized.":[83],"This":[84,128],"holds":[85],"true":[86],"even":[87],"if":[88],"consists":[92],"entirely":[94],"unique":[95],"samples,":[96],"indicating":[97],"deduplication":[99],"at":[100,147],"point":[103],"level":[104],"does":[105],"not":[106],"provide":[107],"meaningful":[109],"privacy":[110],"guarantee.":[111],"Correspondingly":[112],"predict,":[114],"then":[115],"observe,":[116],"delayed":[117],"memorization":[118,168],"for":[119],"fat-tailed":[120],"datasets":[121],"(i.e.,":[122],"those":[123],"with":[124],"more":[125],"samples).":[127],"effect":[129],"is":[130,183],"amplified":[131],"when":[132],"fat-tails":[133],"introduced":[135],"into":[136],"high-level":[137],"rules.":[139],"These":[140],"together":[141],"suggest":[142],"dataset":[144],"diversity,":[145],"particularly":[146],"higher":[148],"levels":[149],"abstraction,":[151],"plays":[152],"role":[155],"staving":[157],"off":[158],"memorization.":[159],"Finally,":[160],"identify":[162],"intermediate":[164],"regime":[165],"partial":[167],"learned":[174],"first":[175],"subsequently":[177],"overproduced":[178],"during":[179],"generation.":[180],"If":[181],"stopped":[184],"this":[186],"regime,":[187],"will":[189],"exhibit":[190],"reversion-to-the-mean":[192],"blandness":[193],"often":[194],"derided":[195],"as":[196],"\"slop\".":[197]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-06-02T00:00:00"}