{"id":"https://openalex.org/W7166649031","doi":"https://doi.org/10.48550/arxiv.2606.29952","title":"Exploiting Local Flatness for Efficient Out-of-Distribution Detection","display_name":"Exploiting Local Flatness for Efficient Out-of-Distribution Detection","publication_year":2026,"publication_date":"2026-06-29","ids":{"openalex":"https://openalex.org/W7166649031","doi":"https://doi.org/10.48550/arxiv.2606.29952"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.29952","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.29952","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":"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"},"type":"preprint","indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2606.29952","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5112276155","display_name":"S. J. Park","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Park, Seonghwan","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139712505","display_name":"Hyunji Jung","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jung, Hyunji","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5033836068","display_name":"D Lee","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lee, Dongyeop","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5058910905","display_name":"N Lee","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lee, Namhoon","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/T10036","display_name":"Advanced Neural Network Applications","score":0.38760000467300415,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/T10036","display_name":"Advanced Neural Network Applications","score":0.38760000467300415,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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":0.23589999973773956,"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":0.10459999740123749,"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/hessian-matrix","display_name":"Hessian matrix","score":0.8022000193595886},{"id":"https://openalex.org/keywords/normalization","display_name":"Normalization (sociology)","score":0.7060999870300293},{"id":"https://openalex.org/keywords/curvature","display_name":"Curvature","score":0.675000011920929},{"id":"https://openalex.org/keywords/flatness","display_name":"Flatness (cosmology)","score":0.5884000062942505},{"id":"https://openalex.org/keywords/detector","display_name":"Detector","score":0.41019999980926514},{"id":"https://openalex.org/keywords/histogram","display_name":"Histogram","score":0.3955000042915344}],"concepts":[{"id":"https://openalex.org/C203616005","wikidata":"https://www.wikidata.org/wiki/Q620495","display_name":"Hessian matrix","level":2,"score":0.8022000193595886},{"id":"https://openalex.org/C136886441","wikidata":"https://www.wikidata.org/wiki/Q926129","display_name":"Normalization (sociology)","level":2,"score":0.7060999870300293},{"id":"https://openalex.org/C195065555","wikidata":"https://www.wikidata.org/wiki/Q214881","display_name":"Curvature","level":2,"score":0.675000011920929},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6071000099182129},{"id":"https://openalex.org/C2778530986","wikidata":"https://www.wikidata.org/wiki/Q5457948","display_name":"Flatness (cosmology)","level":3,"score":0.5884000062942505},{"id":"https://openalex.org/C94915269","wikidata":"https://www.wikidata.org/wiki/Q1834857","display_name":"Detector","level":2,"score":0.41019999980926514},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.40950000286102295},{"id":"https://openalex.org/C53533937","wikidata":"https://www.wikidata.org/wiki/Q185020","display_name":"Histogram","level":3,"score":0.3955000042915344},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.39079999923706055},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.38920000195503235},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3749000132083893},{"id":"https://openalex.org/C52622490","wikidata":"https://www.wikidata.org/wiki/Q1026626","display_name":"Feature extraction","level":2,"score":0.3160000145435333},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.2897000014781952},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.2648000121116638},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.2612999975681305},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.25760000944137573},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.2549999952316284}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.29952","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.29952","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2606.29952","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.29952","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":"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"},"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":{"Detecting":[0],"out-of-distribution":[1],"(OOD)":[2],"data":[3],"is":[4],"crucial":[5],"for":[6,157,209],"reliable":[7],"machine":[8],"learning":[9],"deployment.":[10,212],"Among":[11],"detection":[12],"strategies,":[13],"post-hoc":[14],"methods":[15,48],"are":[16],"particularly":[17],"attractive":[18],"due":[19],"to":[20,42,123,190],"their":[21],"efficiency,":[22],"as":[23],"they":[24],"operate":[25],"directly":[26],"on":[27,55,165],"pre-trained":[28],"networks":[29],"without":[30,160],"requiring":[31,161],"retraining.":[32],"Within":[33],"this":[34,70,79,136],"paradigm,":[35],"one":[36],"promising":[37],"direction":[38],"exploits":[39],"loss-landscape":[40],"curvature":[41,80,90,131],"estimate":[43],"model":[44],"uncertainty;":[45],"however,":[46],"such":[47],"incur":[49],"substantial":[50],"computational":[51,187],"cost":[52],"and":[53,66,82,119,180,199,206],"rely":[54],"implicit":[56],"assumptions":[57],"about":[58],"how":[59],"landscape":[60],"flatness":[61],"differs":[62],"between":[63],"in-distribution":[64],"(ID)":[65],"OOD":[67,85,112,166],"data.":[68,163],"In":[69],"work,":[71],"we":[72,106,141],"provide":[73],"the":[74,95,116,175],"first":[75],"systematic":[76],"investigation":[77],"of":[78],"discrepancy":[81],"show":[83,168],"that":[84,114,149,169],"inputs":[86],"exhibit":[87],"larger":[88],"Hessian":[89,118],"than":[91],"ID":[92,154],"data,":[93],"with":[94],"gap":[96],"widening":[97],"under":[98],"stronger":[99],"distributional":[100],"shifts.":[101],"Motivated":[102],"by":[103,178,183,196],"these":[104],"observations,":[105],"propose":[107],"Fold,":[108],"a":[109,145,191,204],"lightweight":[110],"flatness-modulated":[111],"detector":[113],"leverages":[115],"feature":[117,121],"partial":[120],"normalization":[122,137],"improve":[124],"ID-OOD":[125],"separability":[126],"while":[127,185],"avoiding":[128],"costly":[129],"parameter-space":[130],"approximations.":[132],"To":[133],"optimally":[134],"adapt":[135],"across":[138],"diverse":[139],"datasets,":[140],"further":[142],"introduce":[143],"AutoFold,":[144],"self-supervised":[146],"tuning":[147],"scheme":[148],"synthesizes":[150],"pseudo-OOD":[151],"samples":[152],"via":[153],"logit":[155],"masking":[156],"automatic":[158],"calibration":[159],"external":[162],"Experiments":[164],"benchmarks":[167],"Fold":[170,202],"outperforms":[171],"prior":[172],"methods,":[173],"improving":[174],"average":[176],"AUROC":[177],"1.63%":[179],"reducing":[181],"FPR95":[182],"2.30%,":[184],"maintaining":[186],"efficiency":[188],"comparable":[189],"standard":[192],"forward":[193],"pass.":[194],"Supported":[195],"theoretical":[197],"analysis":[198],"extensive":[200],"ablations,":[201],"provides":[203],"principled":[205],"practical":[207],"solution":[208],"robust":[210],"real-world":[211]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-07-01T00:00:00"}