{"id":"https://openalex.org/W7138945373","doi":"https://doi.org/10.48550/arxiv.2603.16139","title":"Rethinking UMM Visual Generation: Masked Modeling for Efficient Image-Only Pre-training","display_name":"Rethinking UMM Visual Generation: Masked Modeling for Efficient Image-Only Pre-training","publication_year":2026,"publication_date":"2026-03-17","ids":{"openalex":"https://openalex.org/W7138945373","doi":"https://doi.org/10.48550/arxiv.2603.16139"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2603.16139","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.16139","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.2603.16139","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5130190204","display_name":"Peng Sun","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sun, Peng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5130163663","display_name":"Jun Xie","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xie, Jun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5129990033","display_name":"Tao Lin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lin, Tao","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":0.5342000126838684,"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":0.5342000126838684,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.34119999408721924,"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/T13519","display_name":"Historical Architecture and Urbanism","score":0.014499999582767487,"subfield":{"id":"https://openalex.org/subfields/1202","display_name":"History"},"field":{"id":"https://openalex.org/fields/12","display_name":"Arts and Humanities"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.6202999949455261},{"id":"https://openalex.org/keywords/dependency","display_name":"Dependency (UML)","score":0.60589998960495},{"id":"https://openalex.org/keywords/generative-grammar","display_name":"Generative grammar","score":0.5532000064849854},{"id":"https://openalex.org/keywords/component","display_name":"Component (thermodynamics)","score":0.5486000180244446},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.5476999878883362},{"id":"https://openalex.org/keywords/scratch","display_name":"Scratch","score":0.49939998984336853},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.4429999887943268},{"id":"https://openalex.org/keywords/generative-model","display_name":"Generative model","score":0.4377000033855438}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8054999709129333},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.6202999949455261},{"id":"https://openalex.org/C19768560","wikidata":"https://www.wikidata.org/wiki/Q320727","display_name":"Dependency (UML)","level":2,"score":0.60589998960495},{"id":"https://openalex.org/C39890363","wikidata":"https://www.wikidata.org/wiki/Q36108","display_name":"Generative grammar","level":2,"score":0.5532000064849854},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.5486000180244446},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.5476999878883362},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5340999960899353},{"id":"https://openalex.org/C2781235140","wikidata":"https://www.wikidata.org/wiki/Q275131","display_name":"Scratch","level":2,"score":0.49939998984336853},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.4429999887943268},{"id":"https://openalex.org/C167966045","wikidata":"https://www.wikidata.org/wiki/Q5532625","display_name":"Generative model","level":3,"score":0.4377000033855438},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3434000015258789},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.33250001072883606},{"id":"https://openalex.org/C58489278","wikidata":"https://www.wikidata.org/wiki/Q1172284","display_name":"Data set","level":2,"score":0.30959999561309814},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.29899999499320984},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.28700000047683716},{"id":"https://openalex.org/C36464697","wikidata":"https://www.wikidata.org/wiki/Q451553","display_name":"Visualization","level":2,"score":0.27480000257492065},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2606000006198883},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.25940001010894775},{"id":"https://openalex.org/C160920958","wikidata":"https://www.wikidata.org/wiki/Q7662746","display_name":"Synthetic data","level":2,"score":0.25940001010894775},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.2547000050544739},{"id":"https://openalex.org/C169903167","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Test set","level":2,"score":0.2513999938964844},{"id":"https://openalex.org/C115961682","wikidata":"https://www.wikidata.org/wiki/Q860623","display_name":"Image (mathematics)","level":2,"score":0.25040000677108765}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2603.16139","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.16139","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.2603.16139","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2603.16139","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":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Unified":[0],"Multimodal":[1],"Models":[2],"(UMMs)":[3],"are":[4],"often":[5],"constrained":[6],"by":[7],"the":[8,45,67,79,92,150,157],"pre-training":[9,33,160],"of":[10,97,105],"their":[11],"$\\textbf{visual":[12],"generation":[13],"components}$,":[14],"which":[15],"typically":[16],"relies":[17],"on":[18,81,165,169],"inefficient":[19],"paradigms":[20],"and":[21,39,100,113,167,179],"scarce,":[22],"high-quality":[23],"text-image":[24,106],"paired":[25,82],"data.":[26],"In":[27],"this":[28,85],"paper,":[29],"we":[30,51],"systematically":[31],"analyze":[32],"recipes":[34],"for":[35,55],"$\\textbf{UMM":[36],"visual":[37,68],"generation}$":[38],"identify":[40],"these":[41],"two":[42],"issues":[43],"as":[44,174],"major":[46],"bottlenecks.":[47],"To":[48],"address":[49],"them,":[50],"propose":[52],"$\\textbf{Image-Only":[53],"Training":[54],"UMMs":[56],"(IOMM)}$,":[57],"a":[58,95,101],"data-efficient":[59],"two-stage":[60],"training":[61,124],"framework.":[62],"The":[63,88],"first":[64],"stage":[65,90],"pre-trains":[66],"generative":[69,114],"component":[70],"$\\textbf{exclusively}$":[71],"using":[72,94,142],"abundant":[73],"unlabeled":[74,98],"image-only":[75],"data,":[76],"thereby":[77],"removing":[78],"dependency":[80],"data":[83],"$\\textbf{for":[84],"costly":[86],"phase}$.":[87],"second":[89],"fine-tunes":[91],"model":[93,137],"mixture":[96],"images":[99],"small":[102],"curated":[103],"set":[104],"pairs,":[107],"leading":[108],"to":[109,156],"improved":[110],"instruction":[111],"alignment":[112],"quality.":[115],"Extensive":[116],"experiments":[117],"show":[118],"that":[119],"IOMM":[120],"not":[121],"only":[122,143],"improves":[123],"efficiency":[125],"but":[126],"also":[127],"achieves":[128,163],"state-of-the-art":[129],"(SOTA)":[130],"performance.":[131],"For":[132],"example,":[133],"our":[134],"IOMM-B":[135],"(3.6B)":[136],"was":[138],"trained":[139],"from":[140],"scratch":[141],"$\\sim":[144],"\\textbf{1050}$":[145],"H800":[146],"GPU":[147],"hours":[148],"(with":[149],"vast":[151],"majority,":[152],"$\\textbf{1000}$":[153],"hours,":[154],"dedicated":[155],"efficient":[158],"$\\textbf{image-only":[159],"stage}$).":[161],"It":[162],"$\\textbf{0.89}$":[164],"GenEval":[166],"$\\textbf{0.55}$":[168],"WISE--surpassing":[170],"strong":[171],"baselines":[172],"such":[173],"BAGEL-7B":[175],"(0.82":[176],"&":[177,182],"0.55)":[178],"BLIP3-o-4B":[180],"(0.84":[181],"0.50).":[183],"Code":[184],"is":[185],"available":[186],"$\\href{https://github.com/LINs-lab/IOMM}{https://github.com/LINs-lab/IOMM}$.":[187]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-03-20T00:00:00"}