{"id":"https://openalex.org/W7166802441","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1839","title":"EULoInf: Efficient Hessian-Free Entropy Based Uncertainty-Aware Data Influence Approximation","display_name":"EULoInf: Efficient Hessian-Free Entropy Based Uncertainty-Aware Data Influence Approximation","publication_year":2026,"publication_date":"2026-01-01","ids":{"openalex":"https://openalex.org/W7166802441","doi":"https://doi.org/10.18653/v1/2026.findings-acl.1839"},"language":null,"primary_location":{"id":"doi:10.18653/v1/2026.findings-acl.1839","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1839","pdf_url":"https://aclanthology.org/2026.findings-acl.1839.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2026.findings-acl.1839.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5139757850","display_name":"Runxin Cai","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Runxin Cai","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5139743926","display_name":"Jingtan Wang","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jingtan Wang","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5030304400","display_name":"Bryan Kian Hsiang Low","orcid":"https://orcid.org/0000-0003-2808-451X"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Bryan Kian Hsiang Low","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":0.0,"has_fulltext":true,"cited_by_count":0,"citation_normalized_percentile":{"value":0.85768501,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"36911","last_page":"36928"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10928","display_name":"Probabilistic and Robust Engineering Design","score":0.09830000251531601,"subfield":{"id":"https://openalex.org/subfields/1804","display_name":"Statistics, Probability and Uncertainty"},"field":{"id":"https://openalex.org/fields/18","display_name":"Decision Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},"topics":[{"id":"https://openalex.org/T10928","display_name":"Probabilistic and Robust Engineering Design","score":0.09830000251531601,"subfield":{"id":"https://openalex.org/subfields/1804","display_name":"Statistics, Probability and Uncertainty"},"field":{"id":"https://openalex.org/fields/18","display_name":"Decision Sciences"},"domain":{"id":"https://openalex.org/domains/2","display_name":"Social Sciences"}},{"id":"https://openalex.org/T12814","display_name":"Gaussian Processes and Bayesian Inference","score":0.0625,"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/T12261","display_name":"Statistical Mechanics and Entropy","score":0.05510000139474869,"subfield":{"id":"https://openalex.org/subfields/3109","display_name":"Statistical and Nonlinear Physics"},"field":{"id":"https://openalex.org/fields/31","display_name":"Physics and Astronomy"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/entropy","display_name":"Entropy (arrow of time)","score":0.4823000133037567},{"id":"https://openalex.org/keywords/transfer-entropy","display_name":"Transfer entropy","score":0.2605000138282776},{"id":"https://openalex.org/keywords/joint-entropy","display_name":"Joint entropy","score":0.24629999697208405},{"id":"https://openalex.org/keywords/computation","display_name":"Computation","score":0.23100000619888306}],"concepts":[{"id":"https://openalex.org/C106301342","wikidata":"https://www.wikidata.org/wiki/Q4117933","display_name":"Entropy (arrow of time)","level":2,"score":0.4823000133037567},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.43860000371932983},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.40560001134872437},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.36090001463890076},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.35569998621940613},{"id":"https://openalex.org/C121864883","wikidata":"https://www.wikidata.org/wiki/Q677916","display_name":"Statistical physics","level":1,"score":0.2915000021457672},{"id":"https://openalex.org/C182049051","wikidata":"https://www.wikidata.org/wiki/Q17147155","display_name":"Transfer entropy","level":3,"score":0.2605000138282776},{"id":"https://openalex.org/C106752470","wikidata":"https://www.wikidata.org/wiki/Q1364826","display_name":"Joint entropy","level":3,"score":0.24629999697208405},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.23559999465942383},{"id":"https://openalex.org/C45374587","wikidata":"https://www.wikidata.org/wiki/Q12525525","display_name":"Computation","level":2,"score":0.23100000619888306}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2026.findings-acl.1839","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1839","pdf_url":"https://aclanthology.org/2026.findings-acl.1839.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2026.findings-acl.1839","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2026.findings-acl.1839","pdf_url":"https://aclanthology.org/2026.findings-acl.1839.pdf","source":null,"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Findings of the Association for Computational Linguistics: ACL 2026","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320320671","display_name":"National Research Foundation","ror":"https://ror.org/05s0g1g46"},{"id":"https://openalex.org/F4320320696","display_name":"Agency for Science, Technology and Research","ror":"https://ror.org/036wvzt09"},{"id":"https://openalex.org/F4320320709","display_name":"National Research Foundation Singapore","ror":"https://ror.org/03cpyc314"},{"id":"https://openalex.org/F4320334744","display_name":"Institute for Infocomm Research","ror":"https://ror.org/053rfa017"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7166802441.pdf","grobid_xml":"https://content.openalex.org/works/W7166802441.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"In":[0],"Large":[1],"Language":[2],"Model":[3],"post-training,":[4],"highquality":[5],"data":[6,28,74,83,186,196],"effectively":[7,157],"enhances":[8],"model":[9,54,77,164],"performance":[10],"with":[11,52,61,84,91],"fine-tuning,":[12,79],"highlighting":[13],"the":[14,25,69,76,159,174],"need":[15],"to":[16,40,88,126,166],"identify":[17],"high-quality":[18],"and":[19,33,80,148,189,195,202],"beneficial":[20,107],"fine-tuning":[21],"data.However,":[22],"one":[23],"of":[24,71],"most":[26],"popular":[27],"valuation":[29,142,187],"paradigms,":[30],"influence":[31,58,86,145,175],"function":[32],"its":[34,41],"variants,":[35],"is":[36,209],"computationally":[37,140],"expensive":[38],"due":[39],"reliance":[42],"on":[43],"inverse":[44],"Hessian-Vector":[45],"Products":[46],"(iHVP)":[47],"computations":[48],"that":[49,82,114],"scale":[50],"poorly":[51],"increasing":[53],"size.To":[55],"examine":[56],"whether":[57],"values":[59],"correlate":[60],"efficiently":[62],"computable":[63],"intrinsic":[64],"features,":[65],"we":[66,121,132],"empirically":[67],"investigate":[68],"distribution":[70],"top":[72],"influential":[73],"for":[75],"in":[78,163],"observe":[81],"high":[85,92],"tend":[87],"be":[89,105],"those":[90],"predictive":[93],"uncertainty.Yet":[94],"such":[95],"highly":[96],"uncertain":[97],"samples":[98],"exhibit":[99],"a":[100,118,123,139],"dual":[101],"nature,":[102],"which":[103],"can":[104],"either":[106],"or":[108,179],"detrimental":[109],"noisy":[110],"data.Unlike":[111],"traditional":[112],"methods":[113,183],"treat":[115],"uncertainty":[116,147],"as":[117],"standalone":[119],"criterion,":[120],"introduce":[122],"directional":[124],"indicator":[125],"rigorously":[127,169],"disentangle":[128],"these":[129],"opposing":[130],"effects.Formally,":[131],"propose":[133],"EULoInf":[134],"(Entropy-based":[135],"Uncertainty-aware":[136],"Lookahead":[137],"Influence),":[138],"efficient":[141],"framework.By":[143],"approximating":[144],"via":[146],"gradient-based":[149],"validation":[150],"loss":[151],"lookahead,":[152],"EU-LoInf":[153],"avoids":[154],"iHVP":[155],"computation,":[156],"reducing":[158,199],"iHVP-induced":[160],"quadratic":[161],"complexity":[162],"parameters":[165],"linear":[167],"time.We":[168],"derive":[170],"our":[171],"framework":[172],"from":[173],"function.Empirically,":[176],"it":[177],"matches":[178],"even":[180],"outperforms":[181],"prior":[182],"across":[184],"diverse":[185],"tasks":[188],"LLM":[190],"architectures,":[191],"including":[192],"mislabel":[193],"detection":[194],"selection,":[197],"while":[198],"computational":[200],"time":[201],"memory":[203],"usage":[204],"by":[205],"over":[206],"50%.The":[207],"code":[208],"available":[210],"at":[211],"github.com/Rachelcoll/EULoInf.":[212]},"counts_by_year":[],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2026-07-02T00:00:00"}