{"id":"https://openalex.org/W4412886810","doi":"https://doi.org/10.18653/v1/2025.acl-long.93","title":"Binary Classifier Optimization for Large Language Model Alignment","display_name":"Binary Classifier Optimization for Large Language Model Alignment","publication_year":2025,"publication_date":"2025-01-01","ids":{"openalex":"https://openalex.org/W4412886810","doi":"https://doi.org/10.18653/v1/2025.acl-long.93"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2025.acl-long.93","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.93","pdf_url":"https://aclanthology.org/2025.acl-long.93.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":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://aclanthology.org/2025.acl-long.93.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5102082603","display_name":"Seungjae Jung","orcid":null},"institutions":[{"id":"https://openalex.org/I207623266","display_name":"Kao Corporation (Japan)","ror":"https://ror.org/016t1kc57","country_code":"JP","type":"company","lineage":["https://openalex.org/I207623266"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Seungjae Jung","raw_affiliation_strings":["Kakao Corp  LBOX"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Kakao Corp  LBOX","institution_ids":["https://openalex.org/I207623266"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5041553395","display_name":"Gunsoo Han","orcid":null},"institutions":[{"id":"https://openalex.org/I207623266","display_name":"Kao Corporation (Japan)","ror":"https://ror.org/016t1kc57","country_code":"JP","type":"company","lineage":["https://openalex.org/I207623266"]}],"countries":["JP"],"is_corresponding":false,"raw_author_name":"Gunsoo Han","raw_affiliation_strings":["Kakao Corp  LBOX"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Kakao Corp  LBOX","institution_ids":["https://openalex.org/I207623266"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5054971914","display_name":"Daniel Wontae Nam","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Daniel Wontae Nam","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5083499937","display_name":"Kyoung-Woon On","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Kyoung-Woon On","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":3,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1858","last_page":"1872"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10181","display_name":"Natural Language Processing Techniques","score":0.8370000123977661,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.8370000123977661,"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/T10028","display_name":"Topic Modeling","score":0.7506999969482422,"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/T11550","display_name":"Text and Document Classification Technologies","score":0.7174000144004822,"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/computer-science","display_name":"Computer science","score":0.7851966619491577},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.6045489311218262},{"id":"https://openalex.org/keywords/classifier","display_name":"Classifier (UML)","score":0.5359277129173279},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.4389665126800537},{"id":"https://openalex.org/keywords/binary-classification","display_name":"Binary classification","score":0.4177040457725525},{"id":"https://openalex.org/keywords/support-vector-machine","display_name":"Support vector machine","score":0.13013401627540588}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7851966619491577},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.6045489311218262},{"id":"https://openalex.org/C95623464","wikidata":"https://www.wikidata.org/wiki/Q1096149","display_name":"Classifier (UML)","level":2,"score":0.5359277129173279},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.4389665126800537},{"id":"https://openalex.org/C66905080","wikidata":"https://www.wikidata.org/wiki/Q17005494","display_name":"Binary classification","level":3,"score":0.4177040457725525},{"id":"https://openalex.org/C12267149","wikidata":"https://www.wikidata.org/wiki/Q282453","display_name":"Support vector machine","level":2,"score":0.13013401627540588}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2025.acl-long.93","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.93","pdf_url":"https://aclanthology.org/2025.acl-long.93.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":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2025.acl-long.93","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2025.acl-long.93","pdf_url":"https://aclanthology.org/2025.acl-long.93.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":"Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4412886810.pdf","grobid_xml":"https://content.openalex.org/works/W4412886810.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W4391375266","https://openalex.org/W2899084033","https://openalex.org/W2748952813","https://openalex.org/W2390279801","https://openalex.org/W4391913857","https://openalex.org/W2358668433","https://openalex.org/W3204019825","https://openalex.org/W4382934300","https://openalex.org/W2121061354","https://openalex.org/W4285388059"],"abstract_inverted_index":{"In":[0],"real-world":[1],"services":[2],"such":[3,34],"as":[4,35,57,83,106],"ChatGPT,":[5],"aligning":[6],"models":[7],"based":[8],"on":[9,42,47,133,142,148],"user":[10],"feedback":[11],"is":[12],"crucial":[13],"for":[14,110],"improving":[15],"model":[16,160],"performance.However,":[17],"due":[18],"to":[19,181],"the":[20,43,80,89,97,111,121,124,176],"simplicity":[21],"and":[22,54,146,164,171],"convenience":[23],"of":[24,178],"providing":[25],"feedback,":[26],"users":[27],"typically":[28],"offer":[29],"only":[30,72],"basic":[31],"binary":[32,73,77,98,184],"signals,":[33],"'thumbs-up'":[36],"or":[37],"'thumbsdown'.Most":[38],"existing":[39],"alignment":[40,166],"research,":[41],"other":[44],"hand,":[45],"relies":[46],"preference-based":[48],"approaches":[49],"that":[50,67,96],"require":[51],"both":[52],"positive":[53],"negative":[55],"responses":[56],"a":[58,65,76,114,134,149],"pair.We":[59],"propose":[60],"Binary":[61],"Classifier":[62],"Optimization":[63,92],"(BCO),":[64],"technique":[66,118],"effectively":[68,87],"aligns":[69],"LLMs":[70,170],"using":[71],"feedback.BCO":[74],"trains":[75],"classifier,":[78],"where":[79,138],"logit":[81],"serves":[82],"an":[84,107],"implicit":[85],"reward,":[86],"minimizing":[88],"Direct":[90],"Preference":[91],"(DPO)":[93],"loss.We":[94],"demonstrate":[95],"cross-entropy":[99],"loss":[100],"employed":[101],"in":[102,129],"classifier":[103],"training":[104],"acts":[105],"upper":[108],"bound":[109],"DPO":[112],"loss.Additionally,":[113],"novel":[115],"reward":[116],"shift":[117],"further":[119],"minimizes":[120],"gap":[122],"between":[123],"losses.We":[125],"validate":[126],"our":[127,139,179],"methodology":[128],"two":[130],"settings:":[131],"first,":[132],"paired":[135],"preference":[136],"dataset,":[137],"method":[140],"performs":[141],"par":[143],"with":[144],"DPO;":[145],"second,":[147],"Likert-5":[150],"scale":[151],"annotation":[152],"dataset":[153],"which":[154],"stems":[155],"from":[156,183],"real":[157],"users'":[158],"queries.Our":[159],"consistently":[161],"demonstrates":[162],"effective":[163],"robust":[165],"across":[167],"four":[168],"base":[169],"three":[172],"different":[173],"datasets,":[174],"showcasing":[175],"strength":[177],"approach":[180],"learning":[182],"signals.":[185]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":2}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}