{"id":"https://openalex.org/W4415025601","doi":"https://doi.org/10.1016/j.neucom.2026.134636","title":"Noise augmented fine tuning for mitigating hallucinations in large language models","display_name":"Noise augmented fine tuning for mitigating hallucinations in large language models","publication_year":2026,"publication_date":"2026-07-30","ids":{"openalex":"https://openalex.org/W4415025601","doi":"https://doi.org/10.1016/j.neucom.2026.134636"},"language":"en","primary_location":{"id":"doi:10.1016/j.neucom.2026.134636","is_oa":true,"landing_page_url":"https://doi.org/10.1016/j.neucom.2026.134636","pdf_url":null,"source":{"id":"https://openalex.org/S45693802","display_name":"Neurocomputing","issn_l":"0925-2312","issn":["0925-2312","1872-8286"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320990","host_organization_name":"Elsevier BV","host_organization_lineage":["https://openalex.org/P4310320990"],"host_organization_lineage_names":["Elsevier BV"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Neurocomputing","raw_type":"journal-article"},"type":"article","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://doi.org/10.1016/j.neucom.2026.134636","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5007779221","display_name":"Afshin Khadangi","orcid":"https://orcid.org/0000-0002-0496-5219"},"institutions":[{"id":"https://openalex.org/I186903577","display_name":"University of Luxembourg","ror":"https://ror.org/036x5ad56","country_code":"LU","type":"education","lineage":["https://openalex.org/I186903577"]},{"id":"https://openalex.org/I4210134190","display_name":"Recherches Scientifiques Luxembourg","ror":"https://ror.org/0333e3w09","country_code":"LU","type":"facility","lineage":["https://openalex.org/I4210134190"]}],"countries":["LU"],"is_corresponding":true,"raw_author_name":"Afshin Khadangi","raw_affiliation_strings":["SnT, University of Luxembourg, 29 Av. John F. Kennedy, Kirchberg, 1855, Luxembourg"],"raw_orcid":"https://orcid.org/0000-0002-0496-5219","affiliations":[{"raw_affiliation_string":"SnT, University of Luxembourg, 29 Av. John F. Kennedy, Kirchberg, 1855, Luxembourg","institution_ids":["https://openalex.org/I186903577","https://openalex.org/I4210134190"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5084664902","display_name":"Amir Sartipi","orcid":"https://orcid.org/0000-0002-7192-2673"},"institutions":[{"id":"https://openalex.org/I186903577","display_name":"University of Luxembourg","ror":"https://ror.org/036x5ad56","country_code":"LU","type":"education","lineage":["https://openalex.org/I186903577"]},{"id":"https://openalex.org/I4210134190","display_name":"Recherches Scientifiques Luxembourg","ror":"https://ror.org/0333e3w09","country_code":"LU","type":"facility","lineage":["https://openalex.org/I4210134190"]}],"countries":["LU"],"is_corresponding":false,"raw_author_name":"Amir Sartipi","raw_affiliation_strings":["SnT, University of Luxembourg, 29 Av. John F. Kennedy, Kirchberg, 1855, Luxembourg"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"SnT, University of Luxembourg, 29 Av. John F. Kennedy, Kirchberg, 1855, Luxembourg","institution_ids":["https://openalex.org/I186903577","https://openalex.org/I4210134190"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5010688960","display_name":"Igor Tchappi","orcid":"https://orcid.org/0000-0001-5437-1817"},"institutions":[{"id":"https://openalex.org/I186903577","display_name":"University of Luxembourg","ror":"https://ror.org/036x5ad56","country_code":"LU","type":"education","lineage":["https://openalex.org/I186903577"]},{"id":"https://openalex.org/I4210134190","display_name":"Recherches Scientifiques Luxembourg","ror":"https://ror.org/0333e3w09","country_code":"LU","type":"facility","lineage":["https://openalex.org/I4210134190"]}],"countries":["LU"],"is_corresponding":false,"raw_author_name":"Igor Tchappi","raw_affiliation_strings":["SnT, University of Luxembourg, 29 Av. John F. Kennedy, Kirchberg, 1855, Luxembourg"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"SnT, University of Luxembourg, 29 Av. John F. Kennedy, Kirchberg, 1855, Luxembourg","institution_ids":["https://openalex.org/I186903577","https://openalex.org/I4210134190"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5119934601","display_name":"Ramin Bahmani","orcid":null},"institutions":[{"id":"https://openalex.org/I186903577","display_name":"University of Luxembourg","ror":"https://ror.org/036x5ad56","country_code":"LU","type":"education","lineage":["https://openalex.org/I186903577"]},{"id":"https://openalex.org/I4210134190","display_name":"Recherches Scientifiques Luxembourg","ror":"https://ror.org/0333e3w09","country_code":"LU","type":"facility","lineage":["https://openalex.org/I4210134190"]}],"countries":["LU"],"is_corresponding":false,"raw_author_name":"Ramin Bahmani","raw_affiliation_strings":["SnT, University of Luxembourg, 29 Av. John F. Kennedy, Kirchberg, 1855, Luxembourg"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"SnT, University of Luxembourg, 29 Av. John F. Kennedy, Kirchberg, 1855, Luxembourg","institution_ids":["https://openalex.org/I186903577","https://openalex.org/I4210134190"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5007779221"],"corresponding_institution_ids":["https://openalex.org/I186903577","https://openalex.org/I4210134190"],"apc_list":{"value":2470,"currency":"USD","value_usd":2470},"apc_paid":{"value":2470,"currency":"USD","value_usd":2470},"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.01234182,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"703","issue":null,"first_page":"134636","last_page":"134636"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T13702","display_name":"Machine Learning in Healthcare","score":0.9373999834060669,"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/T13702","display_name":"Machine Learning in Healthcare","score":0.9373999834060669,"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/T11512","display_name":"Anomaly Detection Techniques and Applications","score":0.9162999987602234,"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/benchmark","display_name":"Benchmark (surveying)","score":0.710099995136261},{"id":"https://openalex.org/keywords/regularization","display_name":"Regularization (linguistics)","score":0.5928999781608582},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.5820000171661377},{"id":"https://openalex.org/keywords/key","display_name":"Key (lock)","score":0.5412999987602234},{"id":"https://openalex.org/keywords/consistency","display_name":"Consistency (knowledge bases)","score":0.525600016117096},{"id":"https://openalex.org/keywords/source-code","display_name":"Source code","score":0.5218999981880188},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.48399999737739563},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.4643999934196472},{"id":"https://openalex.org/keywords/trustworthiness","display_name":"Trustworthiness","score":0.429500013589859}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7811999917030334},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.710099995136261},{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.5928999781608582},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.5820000171661377},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.5412999987602234},{"id":"https://openalex.org/C2776436953","wikidata":"https://www.wikidata.org/wiki/Q5163215","display_name":"Consistency (knowledge bases)","level":2,"score":0.525600016117096},{"id":"https://openalex.org/C43126263","wikidata":"https://www.wikidata.org/wiki/Q128751","display_name":"Source code","level":2,"score":0.5218999981880188},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.5174000263214111},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.48399999737739563},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4661000072956085},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.4643999934196472},{"id":"https://openalex.org/C153701036","wikidata":"https://www.wikidata.org/wiki/Q659974","display_name":"Trustworthiness","level":2,"score":0.429500013589859},{"id":"https://openalex.org/C165064840","wikidata":"https://www.wikidata.org/wiki/Q1321061","display_name":"Matching (statistics)","level":2,"score":0.42890000343322754},{"id":"https://openalex.org/C29265498","wikidata":"https://www.wikidata.org/wiki/Q7047719","display_name":"Noise measurement","level":3,"score":0.40059998631477356},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.3959999978542328},{"id":"https://openalex.org/C4199805","wikidata":"https://www.wikidata.org/wiki/Q2725903","display_name":"Gaussian noise","level":2,"score":0.3822999894618988},{"id":"https://openalex.org/C163716315","wikidata":"https://www.wikidata.org/wiki/Q901177","display_name":"Gaussian","level":2,"score":0.37279999256134033},{"id":"https://openalex.org/C77618280","wikidata":"https://www.wikidata.org/wiki/Q1155772","display_name":"Scheme (mathematics)","level":2,"score":0.3589000105857849},{"id":"https://openalex.org/C12725497","wikidata":"https://www.wikidata.org/wiki/Q810247","display_name":"Baseline (sea)","level":2,"score":0.3384000062942505},{"id":"https://openalex.org/C61326573","wikidata":"https://www.wikidata.org/wiki/Q1496376","display_name":"Gaussian process","level":3,"score":0.33390000462532043},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.32510000467300415},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.3163999915122986},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.30149999260902405},{"id":"https://openalex.org/C115051666","wikidata":"https://www.wikidata.org/wiki/Q6522493","display_name":"Ranging","level":2,"score":0.2896000146865845},{"id":"https://openalex.org/C2780909371","wikidata":"https://www.wikidata.org/wiki/Q4801092","display_name":"Artificial noise","level":4,"score":0.2784000039100647},{"id":"https://openalex.org/C120936955","wikidata":"https://www.wikidata.org/wiki/Q2155640","display_name":"Empirical research","level":2,"score":0.258899986743927},{"id":"https://openalex.org/C2781170535","wikidata":"https://www.wikidata.org/wiki/Q30587856","display_name":"Noisy data","level":2,"score":0.25699999928474426}],"mesh":[],"locations_count":4,"locations":[{"id":"doi:10.1016/j.neucom.2026.134636","is_oa":true,"landing_page_url":"https://doi.org/10.1016/j.neucom.2026.134636","pdf_url":null,"source":{"id":"https://openalex.org/S45693802","display_name":"Neurocomputing","issn_l":"0925-2312","issn":["0925-2312","1872-8286"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320990","host_organization_name":"Elsevier BV","host_organization_lineage":["https://openalex.org/P4310320990"],"host_organization_lineage_names":["Elsevier BV"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Neurocomputing","raw_type":"journal-article"},{"id":"pmh:oai:arXiv.org:2504.03302","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.03302","pdf_url":"https://arxiv.org/pdf/2504.03302","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":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"pmh:oai:orbilu.uni.lu:10993/68890","is_oa":true,"landing_page_url":"https://orbilu.uni.lu/handle/10993/68890","pdf_url":null,"source":{"id":"https://openalex.org/S4306401815","display_name":"Open Repository and Bibliography (University of Luxembourg)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I186903577","host_organization_name":"University of Luxembourg","host_organization_lineage":["https://openalex.org/I186903577"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by-nc-nd","license_id":"https://openalex.org/licenses/cc-by-nc-nd","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Neurocomputing, 134636 (2026-07)","raw_type":"http://purl.org/coar/resource_type/c_6501"},{"id":"doi:10.48550/arxiv.2504.03302","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2504.03302","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:10.1016/j.neucom.2026.134636","is_oa":true,"landing_page_url":"https://doi.org/10.1016/j.neucom.2026.134636","pdf_url":null,"source":{"id":"https://openalex.org/S45693802","display_name":"Neurocomputing","issn_l":"0925-2312","issn":["0925-2312","1872-8286"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320990","host_organization_name":"Elsevier BV","host_organization_lineage":["https://openalex.org/P4310320990"],"host_organization_lineage_names":["Elsevier BV"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Neurocomputing","raw_type":"journal-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G3285427563","display_name":null,"funder_award_id":"13342933/Gilbert Fridgen","funder_id":"https://openalex.org/F4320321038","funder_display_name":"Fonds National de la Recherche Luxembourg"},{"id":"https://openalex.org/G6326759832","display_name":null,"funder_award_id":"NCER22/IS/16570468/NCER-FT","funder_id":"https://openalex.org/F4320321038","funder_display_name":"Fonds National de la Recherche Luxembourg"}],"funders":[{"id":"https://openalex.org/F4320321038","display_name":"Fonds National de la Recherche Luxembourg","ror":"https://ror.org/039z13y21"},{"id":"https://openalex.org/F4922719017","display_name":"PayPal","ror":null}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Large":[0],"language":[1,151],"models":[2],"(LLMs)":[3],"often":[4,130],"produce":[5],"inaccurate":[6],"or":[7,55,132],"misleading":[8],"content\u2014":[9],"hallucinations":[10],".":[11],"To":[12],"address":[13],"this":[14],"challenge,":[15],"we":[16,66],"introduce":[17],"Noise-Augmented":[18],"Fine-Tuning":[19],"(NoiseFiT)":[20],",":[21,106,174,177,180],"a":[22,60,69,92],"novel":[23],"framework":[24],"that":[25,72,97,124],"leverages":[26],"adaptive":[27,98],"noise":[28,99],"injection":[29,100],"based":[30],"on":[31,117],"the":[32,142,162],"signal-to-noise":[33],"ratio":[34],"(SNR)":[35],"to":[36,81,182],"enhance":[37],"model":[38],"robustness.":[39],"Our":[40],"contribution":[41],"is":[42,101],"threefold.":[43],"First,":[44],"NoiseFiT":[45,125],"selectively":[46],"perturbs":[47],"layers":[48],"identified":[49],"as":[50],"either":[51],"high-SNR":[52],"(more":[53],"robust)":[54],"low-SNR":[56],"(potentially":[57],"under-regularized)":[58],"using":[59],"dynamically":[61],"scaled":[62],"Gaussian":[63],"noise.":[64],"Second,":[65],"further":[67,184],"propose":[68],"hybrid":[70],"loss":[71],"combines":[73],"standard":[74],"cross-entropy,":[75,77],"soft":[76],"and":[78,84,104,120,168,178,187],"consistency":[79],"regularization":[80],"ensure":[82],"stable":[83],"accurate":[85],"outputs":[86],"under":[87],"noisy":[88],"training":[89],"conditions.":[90],"Third,":[91],"theoretical":[93],"analysis":[94],"proposed":[95],"shows":[96],"both":[102],"unbiased":[103],"variance-preserving":[105],"providing":[107],"strong":[108],"guarantees":[109],"for":[110,147],"convergence":[111],"in":[112,136],"expectation.":[113],"Moreover,":[114],"empirical":[115],"results":[116],"multiple":[118],"test":[119],"benchmark":[121,165],"datasets,":[122],"demonstrate":[123],"significantly":[126],"reduces":[127],"hallucination":[128],"rates,":[129],"improving":[131],"matching":[133],"baseline":[134],"performance":[135],"key":[137],"tasks.":[138],"These":[139],"findings":[140],"highlight":[141],"promise":[143],"of":[144],"noise-driven":[145],"strategies":[146],"achieving":[148],"robust,":[149],"trustworthy":[150],"modeling":[152],"without":[153],"incurring":[154],"prohibitive":[155],"computational":[156],"overhead.":[157],"We":[158],"have":[159],"publicly":[160],"released":[161],"fine-tuning":[163],"logs,":[164],"evaluation":[166],"artifacts,":[167],"source":[169],"code":[170],"online":[171],"at":[172],"W&B":[173],"Hugging":[175],"Face":[176],"GitHub":[179],"respectively,":[181],"foster":[183],"research,":[185],"accessibility":[186],"reproducibility.":[188]},"counts_by_year":[],"updated_date":"2026-08-14T07:06:38.338062","created_date":"2025-10-10T00:00:00"}