{"id":"https://openalex.org/W7160236015","doi":"https://doi.org/10.48550/arxiv.2605.00939","title":"From Flat Facts to Sharp Hallucinations: Detecting Stubborn Errors via Gradient Sensitivity","display_name":"From Flat Facts to Sharp Hallucinations: Detecting Stubborn Errors via Gradient Sensitivity","publication_year":2026,"publication_date":"2026-05-01","ids":{"openalex":"https://openalex.org/W7160236015","doi":"https://doi.org/10.48550/arxiv.2605.00939"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2605.00939","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.00939","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.2605.00939","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5128920945","display_name":"Yee Zhing Liew","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liew, Yee Zhing","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5080268176","display_name":"Andrew Huey Ping Tan","orcid":"https://orcid.org/0000-0003-4194-1088"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Tan, Andrew Huey Ping","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5135387383","display_name":"Anwar P. P Abdul Majeed","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Majeed, Anwar P. P. Abdul","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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.2858000099658966,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.2858000099658966,"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/T12536","display_name":"Topological and Geometric Data Analysis","score":0.07689999788999557,"subfield":{"id":"https://openalex.org/subfields/1703","display_name":"Computational Theory and Mathematics"},"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/T12553","display_name":"Psychedelics and Drug Studies","score":0.05829999968409538,"subfield":{"id":"https://openalex.org/subfields/3203","display_name":"Clinical Psychology"},"field":{"id":"https://openalex.org/fields/32","display_name":"Psychology"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/hessian-matrix","display_name":"Hessian matrix","score":0.701200008392334},{"id":"https://openalex.org/keywords/sensitivity","display_name":"Sensitivity (control systems)","score":0.5795999765396118},{"id":"https://openalex.org/keywords/gradient-descent","display_name":"Gradient descent","score":0.4921000003814697},{"id":"https://openalex.org/keywords/noise","display_name":"Noise (video)","score":0.48399999737739563},{"id":"https://openalex.org/keywords/gaussian","display_name":"Gaussian","score":0.44769999384880066},{"id":"https://openalex.org/keywords/gaussian-noise","display_name":"Gaussian noise","score":0.3837999999523163},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.35350000858306885}],"concepts":[{"id":"https://openalex.org/C203616005","wikidata":"https://www.wikidata.org/wiki/Q620495","display_name":"Hessian matrix","level":2,"score":0.701200008392334},{"id":"https://openalex.org/C21200559","wikidata":"https://www.wikidata.org/wiki/Q7451068","display_name":"Sensitivity (control systems)","level":2,"score":0.5795999765396118},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5285999774932861},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.4921000003814697},{"id":"https://openalex.org/C99498987","wikidata":"https://www.wikidata.org/wiki/Q2210247","display_name":"Noise (video)","level":3,"score":0.48399999737739563},{"id":"https://openalex.org/C163716315","wikidata":"https://www.wikidata.org/wiki/Q901177","display_name":"Gaussian","level":2,"score":0.44769999384880066},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.438400000333786},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.435699999332428},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4162999987602234},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.40790000557899475},{"id":"https://openalex.org/C4199805","wikidata":"https://www.wikidata.org/wiki/Q2725903","display_name":"Gaussian noise","level":2,"score":0.3837999999523163},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.35350000858306885},{"id":"https://openalex.org/C2779843651","wikidata":"https://www.wikidata.org/wiki/Q7390335","display_name":"SIGNAL (programming language)","level":2,"score":0.3433000147342682},{"id":"https://openalex.org/C104267543","wikidata":"https://www.wikidata.org/wiki/Q208163","display_name":"Signal processing","level":3,"score":0.2921000123023987},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.29109999537467957},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.2759000062942505},{"id":"https://openalex.org/C100675267","wikidata":"https://www.wikidata.org/wiki/Q1371624","display_name":"Background noise","level":2,"score":0.27459999918937683},{"id":"https://openalex.org/C152361515","wikidata":"https://www.wikidata.org/wiki/Q181328","display_name":"Bernoulli's principle","level":2,"score":0.26989999413490295},{"id":"https://openalex.org/C137270730","wikidata":"https://www.wikidata.org/wiki/Q120811","display_name":"Detection theory","level":3,"score":0.2644999921321869}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2605.00939","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.00939","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.2605.00939","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2605.00939","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":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Traditional":[0],"hallucination":[1],"detection":[2],"fails":[3],"on":[4],"\"Stubborn":[5],"Hallucinations\"":[6],"-":[7],"errors":[8],"where":[9],"LLMs":[10],"are":[11],"confidently":[12],"wrong.":[13],"We":[14,23],"propose":[15],"a":[16,90],"geometric":[17],"solution:":[18],"Embedding-Perturbed":[19],"Gradient":[20],"Sensitivity":[21],"(EPGS).":[22],"hypothesize":[24],"that":[25,81],"while":[26],"robust":[27,91],"facts":[28],"reside":[29],"in":[30,36,59],"flat":[31],"minima,":[32,38],"stubborn":[33],"hallucinations":[34],"sit":[35],"sharp":[37],"supported":[39],"by":[40,47],"brittle":[41],"memorization.":[42,77],"EPGS":[43,82],"detects":[44],"this":[45],"sharpness":[46],"perturbing":[48],"input":[49],"embeddings":[50],"with":[51],"Gaussian":[52],"noise":[53],"and":[54,86],"measuring":[55],"the":[56,69],"resulting":[57],"spike":[58],"gradient":[60],"magnitude.":[61],"This":[62],"acts":[63],"as":[64],"an":[65],"efficient":[66],"proxy":[67],"for":[68,93],"Hessian":[70],"spectrum,":[71],"differentiating":[72],"stable":[73],"knowledge":[74],"from":[75],"unstable":[76],"Our":[78],"experiments":[79],"show":[80],"significantly":[83],"outperforms":[84],"entropy-based":[85],"representation-based":[87],"baselines,":[88],"providing":[89],"signal":[92],"detecting":[94],"high-confidence":[95],"factual":[96],"errors.":[97]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-05-06T00:00:00"}