{"id":"https://openalex.org/W7204885828","doi":"https://doi.org/10.48550/arxiv.2608.30858","title":"GAFT: Geo-Anchored Fine-Tuning for Hazard Identification from Rare Failures","display_name":"GAFT: Geo-Anchored Fine-Tuning for Hazard Identification from Rare Failures","publication_year":2026,"publication_date":"2026-08-31","ids":{"openalex":"https://openalex.org/W7204885828","doi":"https://doi.org/10.48550/arxiv.2608.30858"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2608.30858","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.30858","pdf_url":"https://arxiv.org/pdf/2608.30858","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/2608.30858","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5148937512","display_name":"Yanran Xu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Xu, Yanran","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5075589970","display_name":"Chuanhang Qiu","orcid":"https://orcid.org/0000-0002-3427-4603","observed_orcids":["https://orcid.org/0000-0002-3427-4603"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Qiu, Chuanhang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5148894026","display_name":"Yue Wang","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wang, Yue","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139321361","display_name":"Wenbo Wu","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Wu, Wenbo","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5101786356","display_name":"Zhaoxing Li","orcid":"https://orcid.org/0009-0006-1726-3706","observed_orcids":["https://orcid.org/0009-0006-1726-3706","https://orcid.org/0009-0006-4621-2289"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Li, Zhaoxing","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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.6265000104904175,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.6265000104904175,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.3158999979496002,"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/T11606","display_name":"Infrastructure Maintenance and Monitoring","score":0.004399999976158142,"subfield":{"id":"https://openalex.org/subfields/2205","display_name":"Civil and Structural Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/off-road-navigation","display_name":"off-road navigation","score":0.7390000224113464},{"id":"https://openalex.org/keywords/hazard-identification","display_name":"hazard identification","score":0.7739999890327454},{"id":"https://openalex.org/keywords/rare-failure-events","display_name":"rare failure events","score":0.7559999823570251},{"id":"https://openalex.org/keywords/parameter-efficient-fine-tuning","display_name":"parameter-efficient fine-tuning","score":0.7789999842643738},{"id":"https://openalex.org/keywords/lora","display_name":"LoRa","score":0.9049999713897705},{"id":"https://openalex.org/keywords/vision-foundation-models","display_name":"vision foundation models","score":0.8769999742507935},{"id":"https://openalex.org/keywords/spatial-attention","display_name":"spatial attention","score":0.7540000081062317},{"id":"https://openalex.org/keywords/entrapment","display_name":"entrapment","score":0.9229999780654907},{"id":"https://openalex.org/keywords/dinov2","display_name":"DINOv2","score":0.6890000104904175}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6951000094413757},{"id":"https://openalex.org/C165696696","wikidata":"https://www.wikidata.org/wiki/Q11287","display_name":"Exploit","level":2,"score":0.6317999958992004},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6179999709129333},{"id":"https://openalex.org/C49261128","wikidata":"https://www.wikidata.org/wiki/Q1132455","display_name":"Hazard","level":2,"score":0.5673999786376953},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.5097000002861023},{"id":"https://openalex.org/C116834253","wikidata":"https://www.wikidata.org/wiki/Q2039217","display_name":"Identification (biology)","level":2,"score":0.4993000030517578},{"id":"https://openalex.org/C139807058","wikidata":"https://www.wikidata.org/wiki/Q352374","display_name":"Adaptation (eye)","level":2,"score":0.4726000130176544},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4715000092983246},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.45879998803138733},{"id":"https://openalex.org/C114289077","wikidata":"https://www.wikidata.org/wiki/Q3284399","display_name":"Statistical model","level":2,"score":0.4424000084400177},{"id":"https://openalex.org/C2982736386","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Statistical learning","level":2,"score":0.41100001335144043},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.36039999127388},{"id":"https://openalex.org/C168820333","wikidata":"https://www.wikidata.org/wiki/Q448889","display_name":"Visual inspection","level":2,"score":0.3391999900341034},{"id":"https://openalex.org/C126042441","wikidata":"https://www.wikidata.org/wiki/Q1324888","display_name":"Frame (networking)","level":2,"score":0.33809998631477356},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.30709999799728394},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.3034999966621399},{"id":"https://openalex.org/C124101348","wikidata":"https://www.wikidata.org/wiki/Q172491","display_name":"Data mining","level":1,"score":0.30309998989105225},{"id":"https://openalex.org/C111370547","wikidata":"https://www.wikidata.org/wiki/Q7451120","display_name":"Sensory cue","level":2,"score":0.2863999903202057},{"id":"https://openalex.org/C2777317252","wikidata":"https://www.wikidata.org/wiki/Q18393516","display_name":"Rare events","level":2,"score":0.28200000524520874},{"id":"https://openalex.org/C136389625","wikidata":"https://www.wikidata.org/wiki/Q334384","display_name":"Supervised learning","level":3,"score":0.2777999937534332},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.27140000462532043},{"id":"https://openalex.org/C2986587452","wikidata":"https://www.wikidata.org/wiki/Q938438","display_name":"Statistical analysis","level":2,"score":0.26100000739097595},{"id":"https://openalex.org/C26517878","wikidata":"https://www.wikidata.org/wiki/Q228039","display_name":"Key (lock)","level":2,"score":0.25949999690055847}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2608.30858","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.30858","pdf_url":"https://arxiv.org/pdf/2608.30858","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.2608.30858","doi":"https://doi.org/10.48550/arxiv.2608.30858","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2608.30858","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2608.30858","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2608.30858","pdf_url":"https://arxiv.org/pdf/2608.30858","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/W7222947736","https://openalex.org/W7135609745","https://openalex.org/W7163096005","https://openalex.org/W4416429297","https://openalex.org/W7221569583","https://openalex.org/W7215129353","https://openalex.org/W7104681270","https://openalex.org/W7172449412","https://openalex.org/W7138516009","https://openalex.org/W7214788624"],"abstract_inverted_index":{"Off-road":[0],"navigation":[1],"can":[2,63],"fail":[3],"when":[4],"physical":[5],"structures":[6,20],"induce":[7],"irrecoverable":[8],"states":[9],"such":[10],"as":[11],"high-centering":[12],"or":[13],"entrapment,":[14],"requiring":[15],"human":[16],"interventions.":[17],"Identifying":[18],"these":[19,61,146],"is":[21],"crucial,":[22],"yet":[23],"challenging.":[24],"Such":[25],"failure":[26],"events":[27],"are":[28],"rare":[29],"and":[30,125,161],"costly":[31],"to":[32,70,137],"collect,":[33],"resulting":[34],"in":[35],"limited":[36],"training":[37],"data.":[38],"Moreover,":[39],"the":[40,51,56,102,130,150],"collected":[41],"data":[42,62],"associate":[43],"frames":[44],"with":[45,87,101,139],"outcomes,":[46],"but":[47],"do":[48],"not":[49],"indicate":[50],"visual":[52,67],"cues":[53],"responsible":[54],"for":[55],"failure.":[57],"Learning":[58],"directly":[59],"from":[60,135],"therefore":[64],"exploit":[65],"scenario-specific":[66],"cues,":[68],"leading":[69],"poor":[71],"generalization.":[72],"We":[73],"propose":[74],"\\textbf{Geo-Anchored":[75],"Fine-Tuning":[76],"(GAFT)},":[77],"a":[78,83,88,97,155],"parameter-efficient":[79],"method":[80],"that":[81],"adapts":[82],"vision":[84],"foundation":[85],"model":[86,153],"geometry-derived":[89],"prior.":[90],"It":[91],"guides":[92],"LoRA":[93],"adaptation":[94],"by":[95],"aligning":[96],"spatial":[98],"attention-rollout":[99],"map":[100],"geometry":[103],"prior,":[104],"while":[105],"preserving":[106],"pretrained":[107],"representations.":[108],"On":[109],"an":[110],"intervention-verified":[111],"forest":[112],"hazard":[113],"benchmark,":[114],"across":[115],"ten":[116],"independently":[117,147],"trained":[118,148],"adaptations,":[119],"GAFT":[120,152],"consistently":[121],"outperforms":[122],"frozen":[123],"DINOv2":[124],"supervised":[126],"PEFT":[127],"baselines,":[128],"improving":[129],"repeated":[131],"leave-one-scenario-out":[132],"mean":[133],"$F_2$":[134,157],"0.0607":[136],"0.3757":[138],"statistical":[140],"significance":[141],"under":[142],"paired":[143],"analysis.":[144],"Within":[145],"models,":[149],"best-performing":[151],"achieves":[154],"repeated-LOSO":[156],"of":[158],"0.570.":[159],"Code":[160],"benchmark:":[162],"https://github.com/Xu-Yanran/geo_anchored_fine_tuning":[163]},"counts_by_year":[],"updated_date":"2026-10-28T10:33:24.094632","created_date":"2026-09-02T00:00:00"}