{"id":"https://openalex.org/W7163715307","doi":"https://doi.org/10.48550/arxiv.2606.05414","title":"When Evidence is Sparse: Weakly Supervised Early Failure Alerting in Dialogs and LLM-Agent Trajectories","display_name":"When Evidence is Sparse: Weakly Supervised Early Failure Alerting in Dialogs and LLM-Agent Trajectories","publication_year":2026,"publication_date":"2026-06-03","ids":{"openalex":"https://openalex.org/W7163715307","doi":"https://doi.org/10.48550/arxiv.2606.05414"},"language":null,"primary_location":{"id":"doi:10.48550/arxiv.2606.05414","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.05414","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.2606.05414","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5137971027","display_name":"Avinash Baidya","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Baidya, Avinash","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5086761697","display_name":"Xinran Liang","orcid":"https://orcid.org/0000-0001-8729-7051"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Liang, Xinran","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5054719216","display_name":"Ruocheng Guo","orcid":"https://orcid.org/0000-0002-8522-6142"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Guo, Ruocheng","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5137977039","display_name":"Xiang Gao","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Gao, Xiang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5137952172","display_name":"Kamalika Das","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Das, Kamalika","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/T10028","display_name":"Topic Modeling","score":0.23970000445842743,"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/T10028","display_name":"Topic Modeling","score":0.23970000445842743,"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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.22310000658035278,"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/T12031","display_name":"Speech and dialogue systems","score":0.17579999566078186,"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/inference","display_name":"Inference","score":0.6746000051498413},{"id":"https://openalex.org/keywords/point","display_name":"Point (geometry)","score":0.4596000015735626},{"id":"https://openalex.org/keywords/trajectory","display_name":"Trajectory","score":0.44359999895095825},{"id":"https://openalex.org/keywords/quality","display_name":"Quality (philosophy)","score":0.4359999895095825},{"id":"https://openalex.org/keywords/reliability","display_name":"Reliability (semiconductor)","score":0.3479999899864197},{"id":"https://openalex.org/keywords/bridge","display_name":"Bridge (graph theory)","score":0.3244999945163727},{"id":"https://openalex.org/keywords/encoding","display_name":"Encoding (memory)","score":0.3066999912261963}],"concepts":[{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6746000051498413},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6459000110626221},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.49059998989105225},{"id":"https://openalex.org/C28719098","wikidata":"https://www.wikidata.org/wiki/Q44946","display_name":"Point (geometry)","level":2,"score":0.4596000015735626},{"id":"https://openalex.org/C13662910","wikidata":"https://www.wikidata.org/wiki/Q193139","display_name":"Trajectory","level":2,"score":0.44359999895095825},{"id":"https://openalex.org/C2779530757","wikidata":"https://www.wikidata.org/wiki/Q1207505","display_name":"Quality (philosophy)","level":2,"score":0.4359999895095825},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.39149999618530273},{"id":"https://openalex.org/C43214815","wikidata":"https://www.wikidata.org/wiki/Q7310987","display_name":"Reliability (semiconductor)","level":3,"score":0.3479999899864197},{"id":"https://openalex.org/C100776233","wikidata":"https://www.wikidata.org/wiki/Q2532492","display_name":"Bridge (graph theory)","level":2,"score":0.3244999945163727},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.3066999912261963},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.298799991607666},{"id":"https://openalex.org/C2779662365","wikidata":"https://www.wikidata.org/wiki/Q5416694","display_name":"Event (particle physics)","level":2,"score":0.2955000102519989},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.273499995470047},{"id":"https://openalex.org/C2779664074","wikidata":"https://www.wikidata.org/wiki/Q3518405","display_name":"Terminal (telecommunication)","level":2,"score":0.27090001106262207},{"id":"https://openalex.org/C59656382","wikidata":"https://www.wikidata.org/wiki/Q191536","display_name":"Conjunction (astronomy)","level":2,"score":0.2651999890804291},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.2572000026702881}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2606.05414","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.05414","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.2606.05414","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2606.05414","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":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"Early":[0],"failure":[1,63,82,116,121,131,184],"alerting":[2],"requires":[3],"deciding,":[4],"while":[5,36,231],"a":[6,32,93,143,160],"dialog":[7],"or":[8],"agent":[9],"trajectory":[10,124],"is":[11,23,27,71,83],"still":[12],"unfolding,":[13],"whether":[14],"to":[15,20,56,74,129],"flag":[16],"it":[17,128],"as":[18,31,62],"likely":[19],"fail.":[21],"This":[22],"challenging":[24],"because":[25],"supervision":[26],"typically":[28],"available":[29],"only":[30,187],"trajectory-level":[33],"success/failure":[34],"label":[35,55],"alerts":[37],"must":[38],"be":[39],"raised":[40],"from":[41,98,123,133],"partial":[42,134],"interactions.":[43],"Prior":[44],"early-classification":[45],"methods":[46],"often":[47,86],"bridge":[48],"this":[49,68,89,99,139],"gap":[50],"by":[51,211,225,238],"assigning":[52],"the":[53,105,204,219],"terminal":[54],"every":[57,60],"prefix,":[58],"treating":[59],"turn":[61],"evidence.":[64],"We":[65,136,200],"hypothesize":[66],"that":[67,96,148,182,203,218],"prefix-label":[69],"assumption":[70],"poorly":[72],"matched":[73],"multi-turn":[75],"language":[76],"interactions,":[77],"where":[78],"evidence":[79,101,122,185],"of":[80,189,196,241],"eventual":[81],"sparse":[84,100,119],"and":[85,103,126,177,191,217],"delayed.":[87],"In":[88],"paper,":[90],"we":[91,179],"introduce":[92],"two-stage":[94],"approach":[95],"learns":[97,118],"structure":[102],"uses":[104,127],"resulting":[106],"risk":[107,132],"estimates":[108],"for":[109,163],"controllable":[110],"early":[111],"alerting.":[112],"Specifically,":[113],"our":[114],"attention-based":[115,205],"predictor":[117,140,206],"turn-level":[120],"labels":[125],"estimate":[130],"histories.":[135],"then":[137],"pair":[138],"with":[141],"$\u03b1$-STOP,":[142],"single":[144],"preference-conditioned":[145],"stopping":[146],"policy":[147],"selects":[149],"an":[150],"accuracy-earliness":[151],"operating":[152,236],"point":[153,237],"at":[154],"inference":[155],"time":[156],"rather":[157],"than":[158],"training":[159,233],"separate":[161],"trigger":[162,229],"each":[164],"preference.":[165],"Across":[166],"five":[167],"benchmarks":[168],"spanning":[169],"customer":[170],"support,":[171],"task-oriented":[172],"dialog,":[173],"persuasion,":[174],"tool":[175],"use,":[176],"planning,":[178],"first":[180,192],"show":[181,202],"high-relevance":[183],"occupies":[186],"4.7-11.3%":[188],"turns":[190],"appears":[193],"after":[194],"59.0-83.6\\%":[195],"trajectories":[197],"on":[198],"average.":[199],"further":[201],"improves":[207,222],"Pareto-frontier":[208],"quality":[209,224],"(hypervolume)":[210],"1-10\\%":[212],"over":[213,227],"naive":[214],"prefix":[215],"supervision,":[216],"full":[220],"system":[221],"frontier":[223],"3-42\\%":[226],"state-of-the-art":[228],"policies":[230],"reducing":[232],"cost":[234],"per":[235],"1-3":[239],"orders":[240],"magnitude.":[242]},"counts_by_year":[],"updated_date":"2026-07-28T07:46:37.118299","created_date":"2026-06-06T00:00:00"}