{"id":"https://openalex.org/W7160870629","doi":"https://doi.org/10.48550/arxiv.2605.07488","title":"Efficient Data Selection for Multimodal Models via Incremental Optimization Utility","display_name":"Efficient Data Selection for Multimodal Models via Incremental Optimization Utility","publication_year":2026,"publication_date":"2026-05-08","ids":{"openalex":"https://openalex.org/W7160870629","doi":"https://doi.org/10.48550/arxiv.2605.07488"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.07488","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.07488","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":null,"license_id":null,"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.2605.07488","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5127394635","display_name":"Jinhao Jing","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jing, Jinhao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5098933159","display_name":"Qiannian ZhAO","orcid":"https://orcid.org/0009-0000-5385-8041"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Qiannian","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135890254","display_name":"Chao Huang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Huang, Chao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135905225","display_name":"Zhan Su","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Su, Zhan","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/T12535","display_name":"Machine Learning and Data Classification","score":0.2303999960422516,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.2303999960422516,"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/T11596","display_name":"Constraint Satisfaction and Optimization","score":0.11990000307559967,"subfield":{"id":"https://openalex.org/subfields/1705","display_name":"Computer Networks and Communications"},"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":0.10949999839067459,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/baseline","display_name":"Baseline (sea)","score":0.6888999938964844},{"id":"https://openalex.org/keywords/selection","display_name":"Selection (genetic algorithm)","score":0.5199000239372253},{"id":"https://openalex.org/keywords/heuristic","display_name":"Heuristic","score":0.48500001430511475},{"id":"https://openalex.org/keywords/ranking","display_name":"Ranking (information retrieval)","score":0.47589999437332153},{"id":"https://openalex.org/keywords/point","display_name":"Point (geometry)","score":0.3555000126361847},{"id":"https://openalex.org/keywords/heuristics","display_name":"Heuristics","score":0.35350000858306885},{"id":"https://openalex.org/keywords/reduction","display_name":"Reduction (mathematics)","score":0.33399999141693115},{"id":"https://openalex.org/keywords/bridge","display_name":"Bridge (graph theory)","score":0.33070001006126404}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7556999921798706},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.6888999938964844},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5482000112533569},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.5199000239372253},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5049999952316284},{"id":"https://openalex.org/C173801870","wikidata":"https://www.wikidata.org/wiki/Q201413","display_name":"Heuristic","level":2,"score":0.48500001430511475},{"id":"https://openalex.org/C189430467","wikidata":"https://www.wikidata.org/wiki/Q7293293","display_name":"Ranking (information retrieval)","level":2,"score":0.47589999437332153},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.41280001401901245},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.3555000126361847},{"id":"https://openalex.org/C127705205","wikidata":"https://www.wikidata.org/wiki/Q5748245","display_name":"Heuristics","level":2,"score":0.35350000858306885},{"id":"https://openalex.org/C111335779","wikidata":"https://www.wikidata.org/wiki/Q3454686","display_name":"Reduction (mathematics)","level":2,"score":0.33399999141693115},{"id":"https://openalex.org/C100776233","wikidata":"https://www.wikidata.org/wiki/Q2532492","display_name":"Bridge (graph theory)","level":2,"score":0.33070001006126404},{"id":"https://openalex.org/C68781425","wikidata":"https://www.wikidata.org/wiki/Q2052203","display_name":"Multi-objective optimization","level":2,"score":0.3303999900817871},{"id":"https://openalex.org/C198531522","wikidata":"https://www.wikidata.org/wiki/Q485146","display_name":"Sample (material)","level":2,"score":0.32280001044273376},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.31940001249313354},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.31839999556541443},{"id":"https://openalex.org/C137836250","wikidata":"https://www.wikidata.org/wiki/Q984063","display_name":"Optimization problem","level":2,"score":0.3109999895095825},{"id":"https://openalex.org/C2780735816","wikidata":"https://www.wikidata.org/wiki/Q28324931","display_name":"Incremental learning","level":2,"score":0.2971999943256378},{"id":"https://openalex.org/C2776175482","wikidata":"https://www.wikidata.org/wiki/Q1195816","display_name":"Transfer (computing)","level":2,"score":0.2793000042438507},{"id":"https://openalex.org/C160920958","wikidata":"https://www.wikidata.org/wiki/Q7662746","display_name":"Synthetic data","level":2,"score":0.2750000059604645},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.27489998936653137},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2727000117301941},{"id":"https://openalex.org/C93959086","wikidata":"https://www.wikidata.org/wiki/Q6888345","display_name":"Model selection","level":2,"score":0.25130000710487366}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.07488","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.07488","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2605.07488","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.07488","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":null,"license_id":null,"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":{"The":[0],"scaling":[1],"of":[2,37,61,71],"Large":[3],"Multimodal":[4],"Models":[5],"(LMMs)":[6],"is":[7],"constrained":[8],"by":[9,108,114,122,157],"the":[10,68,85,101,118,135,154,179],"quality-quantity":[11],"trade-off":[12],"inherent":[13],"in":[14,26,184],"synthetic":[15],"data.":[16],"Previous":[17],"approaches,":[18],"such":[19],"as":[20,53],"LLM-as-a-Judge,":[21,144],"have":[22],"proven":[23],"their":[24],"effectiveness":[25],"addressing":[27],"this":[28,41],"but":[29],"suffer":[30],"from":[31,164],"prohibitive":[32],"computational":[33],"costs":[34,107],"and":[35,152],"lack":[36],"interpretability.":[38],"To":[39],"bridge":[40],"gap,":[42],"we":[43],"propose":[44],"One-Step-Train":[45],"(OST),":[46],"a":[47,75,80,127,139],"framework":[48],"that":[49,94],"reformulates":[50],"data":[51],"selection":[52],"an":[54],"incremental":[55],"optimization":[56],"utility":[57,70],"ranking":[58],"problem.":[59],"Instead":[60],"relying":[62],"on":[63,79,84],"semantic":[64],"heuristics,":[65],"OST":[66,95,104],"estimates":[67],"marginal":[69],"each":[72],"sample":[73],"via":[74],"simulated":[76],"single-step":[77],"update":[78],"lightweight":[81],"proxy.":[82],"Experiments":[83],"Qwen":[86],"series":[87],"across":[88],"multimodal":[89],"mathematical":[90],"reasoning":[91,186],"benchmarks":[92],"demonstrate":[93],"achieves":[96,138],"Pareto-optimal":[97],"efficiency.":[98],"By":[99],"selecting":[100],"top-50":[102],"subset,":[103],"reduces":[105],"training":[106],"43%":[109],"(and":[110],"total":[111],"time":[112],"consumption":[113],"17)":[115],"while":[116,161],"surpassing":[117],"strong":[119],"LLM-as-a-Judge":[120],"baseline":[121,156],"1.8":[123],"points.":[124,159],"Furthermore,":[125],"under":[126],"fixed":[128],"compute":[129],"budget,":[130],"our":[131,170],"method":[132],"using":[133],"only":[134],"top-20":[136],"subset":[137],"5.6":[140],"point":[141],"gain":[142],"over":[143],"improves":[145],"upon":[146],"heuristic":[147],"scoring":[148],"baselines":[149],"like":[150],"DEITA,":[151],"outperforms":[153],"Full-SFT":[155,162],"8.8":[158],"Notably,":[160],"suffers":[163],"performance":[165],"degradation":[166],"due":[167],"to":[168],"noise,":[169],"optimization-grounded":[171],"approach":[172],"effectively":[173],"identifies":[174],"toxic":[175],"samples,":[176],"successfully":[177],"reversing":[178],"negative":[180],"transfer":[181],"frequently":[182],"observed":[183],"complex":[185],"tasks.":[187]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-12T00:00:00"}