{"id":"https://openalex.org/W4415976144","doi":"https://doi.org/10.48550/arxiv.2504.02911","title":"Noiser: Bounded Input Perturbations for Attributing Large Language Models","display_name":"Noiser: Bounded Input Perturbations for Attributing Large Language Models","publication_year":2025,"publication_date":"2025-04-03","ids":{"openalex":"https://openalex.org/W4415976144","doi":"https://doi.org/10.48550/arxiv.2504.02911"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2504.02911","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.02911","pdf_url":"https://arxiv.org/pdf/2504.02911","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,"listed_in":[],"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"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2504.02911","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5102590170","display_name":"Mohammad Reza Ghasemi Madani","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Madani, Mohammad Reza Ghasemi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5080213009","display_name":"Aryo Pradipta Gema","orcid":"https://orcid.org/0009-0007-1163-3531","observed_orcids":["https://orcid.org/0009-0007-1163-3531"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gema, Aryo Pradipta","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5128267391","display_name":"Gabriele Sarti","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Sarti, Gabriele","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101669961","display_name":"Yu Zhao","orcid":"https://orcid.org/0000-0002-6489-2226","observed_orcids":["https://orcid.org/0000-0002-6489-2226"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zhao, Yu","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5019106673","display_name":"Pasquale Minervini","orcid":"https://orcid.org/0000-0002-8442-602X","observed_orcids":["https://orcid.org/0000-0002-8442-602X"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Minervini, Pasquale","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5066187890","display_name":"Andrea Passerini","orcid":"https://orcid.org/0000-0002-2765-5395","observed_orcids":["https://orcid.org/0000-0002-2765-5395"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Passerini, Andrea","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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":1.0,"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/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":1.0,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.0,"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":0.0,"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/feature-attribution","display_name":"feature attribution","score":0.9900000095367432},{"id":"https://openalex.org/keywords/large-language-models","display_name":"large language models","score":0.9679999947547913},{"id":"https://openalex.org/keywords/perturbation-based-explanation","display_name":"perturbation-based explanation","score":0.7139999866485596},{"id":"https://openalex.org/keywords/model-robustness","display_name":"model robustness","score":0.7129999995231628},{"id":"https://openalex.org/keywords/explanation-faithfulness","display_name":"explanation fidelity","score":0.7839999794960022},{"id":"https://openalex.org/keywords/post-hoc-explanation","display_name":"post-hoc explanation","score":0.7170000076293945}],"concepts":[{"id":"https://openalex.org/C34388435","wikidata":"https://www.wikidata.org/wiki/Q2267362","display_name":"Bounded function","level":2,"score":0.6819000244140625},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.6639000177383423},{"id":"https://openalex.org/C41608201","wikidata":"https://www.wikidata.org/wiki/Q980509","display_name":"Embedding","level":2,"score":0.5914000272750854},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.52920001745224},{"id":"https://openalex.org/C143299363","wikidata":"https://www.wikidata.org/wiki/Q900584","display_name":"Attribution","level":2,"score":0.478300005197525},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.436599999666214},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.429500013589859},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.4108000099658966},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.3946000039577484},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.365200012922287},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.36419999599456787},{"id":"https://openalex.org/C80444323","wikidata":"https://www.wikidata.org/wiki/Q2878974","display_name":"Theoretical computer science","level":1,"score":0.3614000082015991},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3546000123023987},{"id":"https://openalex.org/C11671645","wikidata":"https://www.wikidata.org/wiki/Q5054567","display_name":"Causal model","level":2,"score":0.33390000462532043},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.3285999894142151},{"id":"https://openalex.org/C2777021972","wikidata":"https://www.wikidata.org/wiki/Q22976830","display_name":"Uniqueness","level":2,"score":0.31709998846054077},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.28540000319480896},{"id":"https://openalex.org/C26760741","wikidata":"https://www.wikidata.org/wiki/Q160402","display_name":"Perception","level":2,"score":0.27869999408721924},{"id":"https://openalex.org/C2781170535","wikidata":"https://www.wikidata.org/wiki/Q30587856","display_name":"Noisy data","level":2,"score":0.2786000072956085},{"id":"https://openalex.org/C122770356","wikidata":"https://www.wikidata.org/wiki/Q1656753","display_name":"Identifiability","level":2,"score":0.2696000039577484}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2504.02911","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.02911","pdf_url":"https://arxiv.org/pdf/2504.02911","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,"listed_in":[],"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":"doi:10.48550/arxiv.2504.02911","doi":"https://doi.org/10.48550/arxiv.2504.02911","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2504.02911","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,"listed_in":[],"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2504.02911","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2504.02911","pdf_url":"https://arxiv.org/pdf/2504.02911","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,"listed_in":[],"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"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W7118382741","https://openalex.org/W7083599158","https://openalex.org/W4415016237","https://openalex.org/W4416036619","https://openalex.org/W7203614400","https://openalex.org/W4391673068","https://openalex.org/W4403752369","https://openalex.org/W7154685440","https://openalex.org/W7166888719","https://openalex.org/W4416035335"],"abstract_inverted_index":{"Feature":[0],"attribution":[1],"(FA)":[2],"methods":[3,115],"are":[4],"common":[5],"post-hoc":[6],"approaches":[7],"that":[8,21,42,72,105],"explain":[9],"how":[10],"Large":[11],"Language":[12],"Models":[13],"(LLMs)":[14],"make":[15],"predictions.":[16,134],"Accordingly,":[17],"generating":[18],"faithful":[19],"attributions":[20],"reflect":[22],"the":[23,28,52,55,63,80,90],"actual":[24],"inner":[25],"behavior":[26],"of":[27,54,118],"model":[29,56,77,133],"is":[30],"crucial.":[31],"In":[32],"this":[33],"paper,":[34],"we":[35,67,103],"introduce":[36],"Noiser,":[37],"a":[38,94,125],"perturbation-based":[39,113],"FA":[40,114],"method":[41],"imposes":[43],"bounded":[44],"noise":[45],"on":[46],"each":[47],"input":[48,60,64],"embedding":[49],"and":[50,100,112,121,127],"measures":[51],"robustness":[53],"against":[57],"partially":[58],"noised":[59],"to":[61,78,82,88],"obtain":[62],"attributions.":[65],"Additionally,":[66],"propose":[68],"an":[69,74],"answerability":[70],"metric":[71],"employs":[73],"instructed":[75],"judge":[76],"assess":[79],"extent":[81],"which":[83],"highly":[84],"scored":[85],"tokens":[86],"suffice":[87],"recover":[89],"predicted":[91],"output.":[92],"Through":[93],"comprehensive":[95],"evaluation":[96],"across":[97],"six":[98],"LLMs":[99],"three":[101],"tasks,":[102],"demonstrate":[104],"Noiser":[106],"consistently":[107],"outperforms":[108],"existing":[109],"gradient-based,":[110],"attention-based,":[111],"in":[116],"terms":[117],"both":[119],"faithfulness":[120],"answerability,":[122],"making":[123],"it":[124],"robust":[126],"effective":[128],"approach":[129],"for":[130],"explaining":[131],"language":[132]},"counts_by_year":[],"updated_date":"2026-09-24T07:57:37.273135","created_date":"2025-10-10T00:00:00"}