{"id":"https://openalex.org/W7160303870","doi":"https://doi.org/10.48550/arxiv.2605.01402","title":"Injecting Distributional Awareness into MLLMs via Reinforcement Learning for Deep Imbalanced Regression","display_name":"Injecting Distributional Awareness into MLLMs via Reinforcement Learning for Deep Imbalanced Regression","publication_year":2026,"publication_date":"2026-05-02","ids":{"openalex":"https://openalex.org/W7160303870","doi":"https://doi.org/10.48550/arxiv.2605.01402"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.01402","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.01402","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.2605.01402","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5135391142","display_name":"Yao Du","orcid":"https://orcid.org/0000-0001-9031-8319"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Du, Yao","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5135336043","display_name":"Shanshan Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Song, Shanshan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135414634","display_name":"Xiaomeng Li","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Xiaomeng","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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.22110000252723694,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.22110000252723694,"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/T10028","display_name":"Topic Modeling","score":0.1315000057220459,"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/T13702","display_name":"Machine Learning in Healthcare","score":0.12039999663829803,"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/reinforcement-learning","display_name":"Reinforcement learning","score":0.704800009727478},{"id":"https://openalex.org/keywords/regression","display_name":"Regression","score":0.6833000183105469},{"id":"https://openalex.org/keywords/suite","display_name":"Suite","score":0.5640000104904175},{"id":"https://openalex.org/keywords/regression-analysis","display_name":"Regression analysis","score":0.5439000129699707},{"id":"https://openalex.org/keywords/concordance","display_name":"Concordance","score":0.44600000977516174},{"id":"https://openalex.org/keywords/supervised-learning","display_name":"Supervised learning","score":0.4278999865055084},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.41620001196861267}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.704800009727478},{"id":"https://openalex.org/C83546350","wikidata":"https://www.wikidata.org/wiki/Q1139051","display_name":"Regression","level":2,"score":0.6833000183105469},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6608999967575073},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.6513000130653381},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5741999745368958},{"id":"https://openalex.org/C79581498","wikidata":"https://www.wikidata.org/wiki/Q1367530","display_name":"Suite","level":2,"score":0.5640000104904175},{"id":"https://openalex.org/C152877465","wikidata":"https://www.wikidata.org/wiki/Q208042","display_name":"Regression analysis","level":2,"score":0.5439000129699707},{"id":"https://openalex.org/C160798450","wikidata":"https://www.wikidata.org/wiki/Q4230870","display_name":"Concordance","level":2,"score":0.44600000977516174},{"id":"https://openalex.org/C136389625","wikidata":"https://www.wikidata.org/wiki/Q334384","display_name":"Supervised learning","level":3,"score":0.4278999865055084},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.41620001196861267},{"id":"https://openalex.org/C117220453","wikidata":"https://www.wikidata.org/wiki/Q5172842","display_name":"Correlation","level":2,"score":0.3831999897956848},{"id":"https://openalex.org/C169258074","wikidata":"https://www.wikidata.org/wiki/Q245748","display_name":"Random forest","level":2,"score":0.3370000123977661},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.3264999985694885},{"id":"https://openalex.org/C48921125","wikidata":"https://www.wikidata.org/wiki/Q10861030","display_name":"Linear regression","level":2,"score":0.32330000400543213},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.29100000858306885},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.2793000042438507},{"id":"https://openalex.org/C167085575","wikidata":"https://www.wikidata.org/wiki/Q6803654","display_name":"Mean squared prediction error","level":2,"score":0.2782999873161316},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.26840001344680786}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.01402","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.01402","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.2605.01402","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.01402","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":[{"id":"https://metadata.un.org/sdg/1","score":0.6766692399978638,"display_name":"No poverty"}],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Multimodal":[0],"large":[1],"language":[2],"models":[3],"(MLLMs)":[4],"struggle":[5],"with":[6,118],"numerical":[7],"regression":[8,19,106,116],"under":[9],"long-tailed":[10,105],"target":[11],"distributions.":[12],"Token-level":[13],"supervised":[14],"fine-tuning":[15],"(SFT)":[16],"and":[17,30,81,89,113,124],"point-wise":[18],"rewards":[20],"bias":[21],"learning":[22,59],"toward":[23],"high-density":[24],"regions,":[25],"leading":[26],"to":[27,78],"regression-to-the-mean":[28],"behavior":[29],"poor":[31],"tail":[32],"performance.":[33],"We":[34],"identify":[35],"the":[36,73],"lack":[37],"of":[38,46,86,104],"cross-sample":[39],"relational":[40],"supervision":[41,71],"as":[42],"a":[43,56,101],"key":[44],"limitation":[45],"existing":[47,114],"MLLM":[48,115],"training":[49],"paradigms.":[50],"To":[51],"address":[52],"it,":[53],"we":[54],"propose":[55],"distribution-aware":[57],"reinforcement":[58],"framework":[60,92],"based":[61],"on":[62,100],"Group":[63],"Relative":[64],"Policy":[65],"Optimization,":[66],"which":[67],"introduces":[68],"batch-level":[69],"comparison-based":[70],"via":[72],"Concordance":[74],"Correlation":[75],"Coefficient-based":[76],"reward":[77],"align":[79],"predicted":[80],"ground-truth":[82],"distributions":[83],"in":[84,122],"terms":[85],"correlation,":[87],"scale,":[88],"mean.":[90],"The":[91],"is":[93],"plug-and-play,":[94],"requiring":[95],"no":[96],"architectural":[97],"modification.":[98],"Experiments":[99],"unified":[102],"suite":[103],"benchmarks":[107],"show":[108],"consistent":[109],"improvements":[110],"over":[111],"SFT":[112],"methods,":[117],"particularly":[119],"strong":[120],"gains":[121],"medium-":[123],"few-shot":[125],"regimes.":[126]},"counts_by_year":[],"updated_date":"2026-07-01T06:00:48.157686","created_date":"2026-05-06T00:00:00"}