{"id":"https://openalex.org/W7165014914","doi":"https://doi.org/10.48550/arxiv.2606.18022","title":"Recursive Scaling in Masked Diffusion Models","display_name":"Recursive Scaling in Masked Diffusion Models","publication_year":2026,"publication_date":"2026-06-16","ids":{"openalex":"https://openalex.org/W7165014914","doi":"https://doi.org/10.48550/arxiv.2606.18022"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.18022","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.18022","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.18022","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137568014","display_name":"Alba Carballo-Castro","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Carballo-Castro, Alba","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138809043","display_name":"Julianna Piskorz","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Piskorz, Julianna","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138778697","display_name":"Paulius Rauba","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Rauba, Paulius","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5138801855","display_name":"Mihaela van der Schaar","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"van der Schaar, Mihaela","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5000947076","display_name":"Pascal Frossard","orcid":"https://orcid.org/0000-0002-4010-714X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Frossard, Pascal","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":false,"primary_topic":{"id":"https://openalex.org/T10775","display_name":"Generative Adversarial Networks and Image Synthesis","score":0.8906000256538391,"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":0.8906000256538391,"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/T11448","display_name":"Face recognition and analysis","score":0.007000000216066837,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.006800000090152025,"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/scaling","display_name":"Scaling","score":0.7645000219345093},{"id":"https://openalex.org/keywords/recursion","display_name":"Recursion (computer science)","score":0.7008000016212463},{"id":"https://openalex.org/keywords/sequence","display_name":"Sequence (biology)","score":0.482699990272522},{"id":"https://openalex.org/keywords/noise-reduction","display_name":"Noise reduction","score":0.4790000021457672},{"id":"https://openalex.org/keywords/diffusion","display_name":"Diffusion","score":0.4684999883174896},{"id":"https://openalex.org/keywords/inference","display_name":"Inference","score":0.4277999997138977},{"id":"https://openalex.org/keywords/reduction","display_name":"Reduction (mathematics)","score":0.34860000014305115}],"concepts":[{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.7645000219345093},{"id":"https://openalex.org/C168773036","wikidata":"https://www.wikidata.org/wiki/Q264164","display_name":"Recursion (computer science)","level":2,"score":0.7008000016212463},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5957000255584717},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.4862000048160553},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.482699990272522},{"id":"https://openalex.org/C163294075","wikidata":"https://www.wikidata.org/wiki/Q581861","display_name":"Noise reduction","level":2,"score":0.4790000021457672},{"id":"https://openalex.org/C69357855","wikidata":"https://www.wikidata.org/wiki/Q163214","display_name":"Diffusion","level":2,"score":0.4684999883174896},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.45649999380111694},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.4277999997138977},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.3887999951839447},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.34860000014305115},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.3456999957561493},{"id":"https://openalex.org/C70518719","wikidata":"https://www.wikidata.org/wiki/Q16928317","display_name":"Recursive filter","level":5,"score":0.3142000138759613},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.31060001254081726},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.29179999232292175},{"id":"https://openalex.org/C71559656","wikidata":"https://www.wikidata.org/wiki/Q671298","display_name":"Divide and conquer algorithms","level":2,"score":0.26739999651908875},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.2669000029563904},{"id":"https://openalex.org/C2982939207","wikidata":"https://www.wikidata.org/wiki/Q1175101","display_name":"Recursive functions","level":2,"score":0.26249998807907104},{"id":"https://openalex.org/C121864883","wikidata":"https://www.wikidata.org/wiki/Q677916","display_name":"Statistical physics","level":1,"score":0.258899986743927},{"id":"https://openalex.org/C167928553","wikidata":"https://www.wikidata.org/wiki/Q1376021","display_name":"Estimation theory","level":2,"score":0.2538999915122986}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.18022","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.18022","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.18022","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.18022","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"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"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Masked":[0,33],"diffusion":[1,55],"models":[2,122],"(MDMs)":[3],"have":[4],"recently":[5],"emerged":[6],"as":[7,41],"a":[8,42,92,143],"promising":[9],"paradigm":[10],"for":[11,116,148],"sequence":[12],"generation.":[13],"Scaling":[14],"MDMs":[15],"is":[16,142],"conventionally":[17],"achieved":[18],"by":[19,46],"increasing":[20,67,72],"the":[21,25,49,62,100,125,155],"parameter":[22,65,73,90,152],"count":[23],"or":[24],"number":[26],"of":[27,61,102,157],"denoising":[28,51,118],"steps.":[29],"We":[30],"introduce":[31],"Recursive":[32],"Diffusion":[34],"Models":[35],"(R-MDMs),":[36],"which":[37],"add":[38],"recursive":[39,96,111,121,140],"depth":[40,70,141],"third":[43],"scaling":[44,146],"axis":[45],"repeatedly":[47],"applying":[48],"same":[50,126],"transformer":[52],"within":[53],"each":[54],"step.":[56],"Recursion":[57],"enables":[58],"iterative":[59],"refinement":[60,112],"output":[63],"through":[64],"reuse,":[66],"effective":[68],"model":[69,93],"without":[71],"count.":[74],"Across":[75],"structured":[76],"generation":[77,127],"tasks,":[78],"including":[79],"Sudoku":[80],"and":[81,154],"Countdown,":[82],"we":[83],"show":[84],"that":[85,139],"R-MDMs":[86],"achieve":[87],"substantially":[88],"improved":[89],"efficiency:":[91],"with":[94,105,129],"$L$":[95],"iterations":[97],"often":[98],"matches":[99],"performance":[101],"non-recursive":[103],"baselines":[104],"roughly":[106],"$L\\times$":[107],"more":[108],"parameters.":[109],"Moreover,":[110],"can":[113],"partially":[114],"substitute":[115],"additional":[117],"steps,":[119],"allowing":[120],"to":[123],"reach":[124],"quality":[128],"fewer":[130],"forward":[131],"passes":[132],"at":[133],"inference":[134],"time.":[135],"These":[136],"results":[137],"suggest":[138],"practically":[144],"useful":[145],"mechanism":[147],"MDMs,":[149],"improving":[150],"both":[151],"efficiency":[153],"allocation":[156],"test-time":[158],"compute.":[159]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-06-18T00:00:00"}