{"id":"https://openalex.org/W7164037123","doi":"https://doi.org/10.48550/arxiv.2606.09466","title":"DECSELFMASK: Leveraging Unlabeled Text via Self-Relevance-Guided Masking for Decoder-Only Classification","display_name":"DECSELFMASK: Leveraging Unlabeled Text via Self-Relevance-Guided Masking for Decoder-Only Classification","publication_year":2026,"publication_date":"2026-06-08","ids":{"openalex":"https://openalex.org/W7164037123","doi":"https://doi.org/10.48550/arxiv.2606.09466"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.09466","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.09466","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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.2606.09466","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5116866766","display_name":"Pietro Ferrazzi","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ferrazzi, Pietro","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113392231","display_name":"Matteo Merler","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Merler, Matteo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5015514303","display_name":"Giovanni Bonetta","orcid":"https://orcid.org/0000-0003-4498-1026"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bonetta, Giovanni","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5073069826","display_name":"Alberto Lavelli","orcid":"https://orcid.org/0000-0002-7175-6804"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lavelli, Alberto","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5138218663","display_name":"Bernardo Magnini","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Magnini, Bernardo","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/T10028","display_name":"Topic Modeling","score":0.2451000064611435,"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/T10028","display_name":"Topic Modeling","score":0.2451000064611435,"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/T11550","display_name":"Text and Document Classification Technologies","score":0.23960000276565552,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.07410000264644623,"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/masking","display_name":"Masking (illustration)","score":0.7511000037193298},{"id":"https://openalex.org/keywords/semantics","display_name":"Semantics (computer science)","score":0.5551000237464905},{"id":"https://openalex.org/keywords/relevance","display_name":"Relevance (law)","score":0.542900025844574},{"id":"https://openalex.org/keywords/training-set","display_name":"Training set","score":0.5206999778747559},{"id":"https://openalex.org/keywords/labeled-data","display_name":"Labeled data","score":0.4456999897956848},{"id":"https://openalex.org/keywords/synthetic-data","display_name":"Synthetic data","score":0.36320000886917114},{"id":"https://openalex.org/keywords/downstream","display_name":"Downstream (manufacturing)","score":0.349700003862381},{"id":"https://openalex.org/keywords/statistical-model","display_name":"Statistical model","score":0.30809998512268066}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8223000168800354},{"id":"https://openalex.org/C2777402240","wikidata":"https://www.wikidata.org/wiki/Q6783436","display_name":"Masking (illustration)","level":2,"score":0.7511000037193298},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6072999835014343},{"id":"https://openalex.org/C184337299","wikidata":"https://www.wikidata.org/wiki/Q1437428","display_name":"Semantics (computer science)","level":2,"score":0.5551000237464905},{"id":"https://openalex.org/C158154518","wikidata":"https://www.wikidata.org/wiki/Q7310970","display_name":"Relevance (law)","level":2,"score":0.542900025844574},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.5206999778747559},{"id":"https://openalex.org/C2776145971","wikidata":"https://www.wikidata.org/wiki/Q30673951","display_name":"Labeled data","level":2,"score":0.4456999897956848},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4401000142097473},{"id":"https://openalex.org/C160920958","wikidata":"https://www.wikidata.org/wiki/Q7662746","display_name":"Synthetic data","level":2,"score":0.36320000886917114},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.35519999265670776},{"id":"https://openalex.org/C2776207758","wikidata":"https://www.wikidata.org/wiki/Q5303302","display_name":"Downstream (manufacturing)","level":2,"score":0.349700003862381},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.30809998512268066},{"id":"https://openalex.org/C16910744","wikidata":"https://www.wikidata.org/wiki/Q7705759","display_name":"Test data","level":2,"score":0.30410000681877136},{"id":"https://openalex.org/C2781170535","wikidata":"https://www.wikidata.org/wiki/Q30587856","display_name":"Noisy data","level":2,"score":0.2948000133037567},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.28130000829696655},{"id":"https://openalex.org/C136389625","wikidata":"https://www.wikidata.org/wiki/Q334384","display_name":"Supervised learning","level":3,"score":0.27869999408721924},{"id":"https://openalex.org/C166955791","wikidata":"https://www.wikidata.org/wiki/Q629579","display_name":"Macro","level":2,"score":0.27619999647140503},{"id":"https://openalex.org/C108583219","wikidata":"https://www.wikidata.org/wiki/Q197536","display_name":"Deep learning","level":2,"score":0.2653000056743622},{"id":"https://openalex.org/C67186912","wikidata":"https://www.wikidata.org/wiki/Q367664","display_name":"Data modeling","level":2,"score":0.26260000467300415},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.25529998540878296},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.2549000084400177}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.09466","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.09466","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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.2606.09466","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.09466","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"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":{"Classification":[0],"tasks":[1,139,157],"require":[2],"annotated":[3,30],"data,":[4],"which":[5],"can":[6,127],"often":[7,27],"be":[8,128],"expensive,":[9],"time-consuming,":[10],"or":[11],"even":[12],"unfeasible":[13],"to":[14,44,61,68,80,106],"collect.":[15],"This":[16],"is":[17],"the":[18,21,104,119],"case":[19],"of":[20,84,123,143,161],"medical":[22],"domain,":[23],"where":[24],"large":[25],"datasets":[26],"have":[28],"few":[29],"examples.":[31],"To":[32],"address":[33],"this,":[34],"we":[35],"propose":[36,69],"DecSelfMask":[37],"(Decoder":[38],"Self-learning":[39],"by":[40,57,98],"Masking),":[41],"an":[42,148],"approach":[43,136],"enhance":[45],"decoder-only":[46],"performance":[47],"on":[48,53,137,155,158],"classification":[49],"tasks.":[50],"We":[51,75,92,111,133,151],"build":[52],"common":[54,195],"self-learning":[55],"approaches":[56,178],"leveraging":[58],"a":[59,70,90,141,167],"model":[60,105],"create":[62,94],"training":[63,96,103],"examples":[64,97,115],"from":[65,140,147],"unlabeled":[66],"data":[67,125],"novel":[71],"relevance-guided":[72],"masking":[73,99],"strategy.":[74],"use":[76],"relevance":[77],"attribution":[78],"methods":[79],"determine":[81],"what":[82],"portions":[83],"unannotated":[85,124],"texts":[86],"are":[87],"relevant":[88],"for":[89,130],"task.":[91],"then":[93],"self-supervised":[95],"out":[100],"those":[101,114],"portions,":[102],"reconstruct":[107],"them":[108],"via":[109],"next-token-prediction.":[110],"hypothesize":[112],"that":[113,126],"convey":[116],"knowledge":[117],"about":[118],"structure":[120],"and":[121,164,188],"semantics":[122],"useful":[129],"downstream":[131,156],"performance.":[132],"test":[134],"our":[135],"136":[138],"collection":[142],"1.9M":[144],"clinical":[145],"notes":[146],"Italian":[149],"hospital.":[150],"quantify":[152],"DecSelfMask's":[153],"impact":[154],"5":[159],"models":[160],"different":[162],"scales":[163],"families,":[165],"including":[166],"probing":[168],"analysis.":[169],"Experiments":[170],"show":[171],"consistent":[172],"gains,":[173],"outperforming":[174],"standard":[175],"supervised":[176],"fine-tuning":[177],"(+19.9":[179],"points":[180],"in":[181],"Macro":[182],"F1),":[183],"synthetic":[184],"label":[185],"generation":[186],"(+12.5),":[187],"continual":[189],"pretraining":[190],"(+6.3),":[191],"as":[192,194],"well":[193],"baselines.":[196]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-10T00:00:00"}