{"id":"https://openalex.org/W7159688438","doi":"https://doi.org/10.48550/arxiv.2604.27932","title":"Dynamic Cluster Data Sampling for Efficient and Long-Tail-Aware Vision-Language Pre-training","display_name":"Dynamic Cluster Data Sampling for Efficient and Long-Tail-Aware Vision-Language Pre-training","publication_year":2026,"publication_date":"2026-04-30","ids":{"openalex":"https://openalex.org/W7159688438","doi":"https://doi.org/10.48550/arxiv.2604.27932"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2604.27932","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.27932","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.2604.27932","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5134962807","display_name":"Mingliang Liang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liang, Mingliang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5011975959","display_name":"Zhuoran Liu","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liu, Zhuoran","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5060259737","display_name":"Arjen P. de Vries","orcid":"https://orcid.org/0000-0002-2888-4202"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"de Vries, Arjen P.","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5134935271","display_name":"Martha Larson","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Larson, Martha","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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9732999801635742,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9732999801635742,"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":0.004600000102072954,"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/T10627","display_name":"Advanced Image and Video Retrieval Techniques","score":0.00279999990016222,"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/sampling","display_name":"Sampling (signal processing)","score":0.6797999739646912},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.48159998655319214},{"id":"https://openalex.org/keywords/flattening","display_name":"Flattening","score":0.44620001316070557},{"id":"https://openalex.org/keywords/importance-sampling","display_name":"Importance sampling","score":0.4174000024795532},{"id":"https://openalex.org/keywords/distribution","display_name":"Distribution (mathematics)","score":0.3758000135421753},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.36739999055862427},{"id":"https://openalex.org/keywords/dynamic-data","display_name":"Dynamic data","score":0.35850000381469727},{"id":"https://openalex.org/keywords/cluster-sampling","display_name":"Cluster sampling","score":0.32199999690055847}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7522000074386597},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.6797999739646912},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.48159998655319214},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.45980000495910645},{"id":"https://openalex.org/C19444555","wikidata":"https://www.wikidata.org/wiki/Q212750","display_name":"Flattening","level":2,"score":0.44620001316070557},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.4174000024795532},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3937999904155731},{"id":"https://openalex.org/C110121322","wikidata":"https://www.wikidata.org/wiki/Q865811","display_name":"Distribution (mathematics)","level":2,"score":0.3758000135421753},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.36899998784065247},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.36739999055862427},{"id":"https://openalex.org/C197298091","wikidata":"https://www.wikidata.org/wiki/Q5318963","display_name":"Dynamic data","level":2,"score":0.35850000381469727},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3497999906539917},{"id":"https://openalex.org/C183380357","wikidata":"https://www.wikidata.org/wiki/Q1776598","display_name":"Cluster sampling","level":3,"score":0.32199999690055847},{"id":"https://openalex.org/C182306322","wikidata":"https://www.wikidata.org/wiki/Q1779371","display_name":"Order (exchange)","level":2,"score":0.30140000581741333},{"id":"https://openalex.org/C18762648","wikidata":"https://www.wikidata.org/wiki/Q42213","display_name":"Work (physics)","level":2,"score":0.3010999858379364},{"id":"https://openalex.org/C164866538","wikidata":"https://www.wikidata.org/wiki/Q367351","display_name":"Cluster (spacecraft)","level":2,"score":0.2985000014305115},{"id":"https://openalex.org/C167723999","wikidata":"https://www.wikidata.org/wiki/Q3773214","display_name":"Sampling distribution","level":2,"score":0.28859999775886536},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.2822999954223633},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.2752000093460083},{"id":"https://openalex.org/C9652623","wikidata":"https://www.wikidata.org/wiki/Q190109","display_name":"Field (mathematics)","level":2,"score":0.2662999927997589},{"id":"https://openalex.org/C2779343474","wikidata":"https://www.wikidata.org/wiki/Q3109175","display_name":"Context (archaeology)","level":2,"score":0.2662000060081482},{"id":"https://openalex.org/C2985946229","wikidata":"https://www.wikidata.org/wiki/Q49908","display_name":"Data sampling","level":2,"score":0.2526000142097473}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2604.27932","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.27932","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.2604.27932","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2604.27932","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":{"The":[0],"computational":[1,169],"cost":[2,170],"of":[3,28,35,92,103,112,136,144,171],"training":[4,15,56,68,173],"a":[5,59,82,176],"vision-language":[6],"model":[7],"(VLM)":[8],"can":[9],"be":[10],"reduced":[11],"by":[12],"sampling":[13,85,149],"the":[14,26,33,38,55,67,101,109,116,120,133,137,142,168],"data.":[16,138],"Previous":[17],"work":[18],"on":[19,131],"efficient":[20,46],"VLM":[21,42,172],"pre-training":[22,47],"has":[23],"pointed":[24],"to":[25,40,153],"importance":[27,143],"semantic":[29,113,134],"data":[30,39,69,93,117,156],"balance,":[31],"adjusting":[32],"distribution":[34,135],"topics":[36],"in":[37,66,115],"improve":[41,154],"accuracy.":[43],"However,":[44],"existing":[45],"approaches":[48],"may":[49],"disproportionately":[50],"remove":[51],"rare":[52],"concepts":[53,62],"from":[54],"corpus.":[57],"As":[58],"result,":[60],"long-tail":[61,180],"remain":[63],"insufficiently":[64],"represented":[65],"and":[70,94,118,158,174],"are":[71],"not":[72],"effectively":[73],"captured":[74],"during":[75],"training.":[76],"In":[77],"this":[78],"work,":[79,127],"we":[80,140],"introduce":[81],"dynamic":[83,145],"cluster-based":[84],"approach":[86,123],"(DynamiCS)":[87],"that":[88,165],"downsamples":[89],"large":[90],"clusters":[91,114],"upsamples":[95],"small":[96],"ones.":[97],"We":[98],"first":[99],"demonstrate":[100],"advantage":[102,178],"our":[104],"cluster-scaling":[105],"approach,":[106],"which":[107,128,147],"maintains":[108],"relative":[110],"order":[111],"emphasizes":[119],"long-tail.":[121],"This":[122],"contrasts":[124],"with":[125],"current":[126],"focuses":[129],"only":[130],"flattening":[132],"Then,":[139],"show":[141,164],"sampling,":[146],"applies":[148],"at":[150,184],"each":[151],"epoch":[152],"cross-epoch":[155],"diversity":[157],"make":[159],"upsampling":[160],"practical.":[161],"Our":[162],"experiments":[163],"DynamiCS":[166],"reduces":[167],"provides":[175],"performance":[177],"for":[179],"concepts.":[181],"Code":[182],"available":[183],"https://github.com/MingliangLiang3/DynamiCS.":[185]},"counts_by_year":[],"updated_date":"2026-07-09T05:49:46.723101","created_date":"2026-05-02T00:00:00"}