{"id":"https://openalex.org/W7214928710","doi":"https://doi.org/10.48550/arxiv.2609.37209","title":"Pointwise or Pairwise: When Do Pairwise Losses Help Reward Learning, Provably?","display_name":"Pointwise or Pairwise: When Do Pairwise Losses Help Reward Learning, Provably?","publication_year":2026,"publication_date":"2026-09-29","ids":{"openalex":"https://openalex.org/W7214928710","doi":"https://doi.org/10.48550/arxiv.2609.37209"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2609.37209","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.37209","pdf_url":"https://arxiv.org/pdf/2609.37209","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","is_traditional":true,"indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2609.37209","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5153674856","display_name":"Junghyun Lee","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Lee, Junghyun","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5153675161","display_name":"Minsoo Ha","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]},{"id":"https://openalex.org/I4405259277","display_name":"Samsung Electronics (South Korea)","ror":"https://ror.org/020m7t761","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973","https://openalex.org/I4405259277"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Ha, Minsoo","raw_affiliation_strings":["Samsung Research Seoul 06765 , Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Samsung Research Seoul 06765 , Republic of Korea","institution_ids":["https://openalex.org/I2250650973","https://openalex.org/I4405259277"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5125737518","display_name":"Sanghwa Kim","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I123900574","display_name":"Pohang University of Science and Technology","ror":"https://ror.org/04xysgw12","country_code":"KR","type":"education","lineage":["https://openalex.org/I123900574"]},{"id":"https://openalex.org/I157485424","display_name":"Korea Advanced Institute of Science and Technology","ror":"https://ror.org/05apxxy63","country_code":"KR","type":"education","lineage":["https://openalex.org/I157485424"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Kim, Sanghwa","raw_affiliation_strings":["Graduate School of AI , POSTECH Pohang 37673 , Republic of Korea","Kim Jaechul Graduate School of AI , KAIST Seoul 02455 , Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Graduate School of AI , POSTECH Pohang 37673 , Republic of Korea","institution_ids":["https://openalex.org/I123900574"]},{"raw_affiliation_string":"Kim Jaechul Graduate School of AI , KAIST Seoul 02455 , Republic of Korea","institution_ids":["https://openalex.org/I157485424"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5153668672","display_name":"Yeongjong Kim","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I123900574","display_name":"Pohang University of Science and Technology","ror":"https://ror.org/04xysgw12","country_code":"KR","type":"education","lineage":["https://openalex.org/I123900574"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Kim, Yeongjong","raw_affiliation_strings":["Graduate School of AI , POSTECH Pohang 37673 , Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Graduate School of AI , POSTECH Pohang 37673 , Republic of Korea","institution_ids":["https://openalex.org/I123900574"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5153620622","display_name":"Eunjee Lee","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I123900574","display_name":"Pohang University of Science and Technology","ror":"https://ror.org/04xysgw12","country_code":"KR","type":"education","lineage":["https://openalex.org/I123900574"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Lee, Eunjee","raw_affiliation_strings":["Graduate School of AI , POSTECH Pohang 37673 , Republic of Korea","Independent Researcher Cambridge , MA , USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Graduate School of AI , POSTECH Pohang 37673 , Republic of Korea","institution_ids":["https://openalex.org/I123900574"]},{"raw_affiliation_string":"Independent Researcher Cambridge , MA , USA","institution_ids":[]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5039690995","display_name":"Seiyun Shin","orcid":"https://orcid.org/0000-0001-9906-8109","observed_orcids":["https://orcid.org/0000-0001-9906-8109"]},"institutions":[{"id":"https://openalex.org/I123900574","display_name":"Pohang University of Science and Technology","ror":"https://ror.org/04xysgw12","country_code":"KR","type":"education","lineage":["https://openalex.org/I123900574"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Shin, Seiyun","raw_affiliation_strings":["Graduate School of AI , POSTECH Pohang 37673 , Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Graduate School of AI , POSTECH Pohang 37673 , Republic of Korea","institution_ids":["https://openalex.org/I123900574"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5086952864","display_name":"Jun Kwang-Sung","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I123900574","display_name":"Pohang University of Science and Technology","ror":"https://ror.org/04xysgw12","country_code":"KR","type":"education","lineage":["https://openalex.org/I123900574"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Jun, Kwang-Sung","raw_affiliation_strings":["Graduate School of AI , POSTECH Pohang 37673 , Republic of Korea"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Graduate School of AI , POSTECH Pohang 37673 , Republic of Korea","institution_ids":["https://openalex.org/I123900574"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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/T12101","display_name":"Advanced Bandit Algorithms Research","score":0.9994999766349792,"subfield":{"id":"https://openalex.org/subfields/1803","display_name":"Management Science and Operations Research"},"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/T12101","display_name":"Advanced Bandit Algorithms Research","score":0.9994999766349792,"subfield":{"id":"https://openalex.org/subfields/1803","display_name":"Management Science and Operations Research"},"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.00019999999494757503,"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/T12072","display_name":"Machine Learning and Algorithms","score":0.0,"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/reward-learning","display_name":"reward learning","score":0.9369999766349792},{"id":"https://openalex.org/keywords/pairwise-loss","display_name":"pairwise loss","score":0.9660000205039978},{"id":"https://openalex.org/keywords/bias-variance-trade-off","display_name":"bias-variance trade-off","score":0.8999999761581421},{"id":"https://openalex.org/keywords/semiparametric-models","display_name":"semiparametric models","score":0.7390000224113464}],"concepts":[{"id":"https://openalex.org/C2777984123","wikidata":"https://www.wikidata.org/wiki/Q9248237","display_name":"Pointwise","level":2,"score":0.9541000127792358},{"id":"https://openalex.org/C184898388","wikidata":"https://www.wikidata.org/wiki/Q1435712","display_name":"Pairwise comparison","level":2,"score":0.7874000072479248},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.6947000026702881},{"id":"https://openalex.org/C61797465","wikidata":"https://www.wikidata.org/wiki/Q1188986","display_name":"Term (time)","level":2,"score":0.522599995136261},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.4934999942779541},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.4912000000476837},{"id":"https://openalex.org/C48921125","wikidata":"https://www.wikidata.org/wiki/Q10861030","display_name":"Linear regression","level":2,"score":0.47099998593330383},{"id":"https://openalex.org/C196083921","wikidata":"https://www.wikidata.org/wiki/Q7915758","display_name":"Variance (accounting)","level":2,"score":0.4652999937534332},{"id":"https://openalex.org/C83546350","wikidata":"https://www.wikidata.org/wiki/Q1139051","display_name":"Regression","level":2,"score":0.4318999946117401},{"id":"https://openalex.org/C2776291640","wikidata":"https://www.wikidata.org/wiki/Q2912517","display_name":"Value (mathematics)","level":2,"score":0.42649999260902405},{"id":"https://openalex.org/C152877465","wikidata":"https://www.wikidata.org/wiki/Q208042","display_name":"Regression analysis","level":2,"score":0.4020000100135803},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.4018000066280365},{"id":"https://openalex.org/C2780009758","wikidata":"https://www.wikidata.org/wiki/Q6804172","display_name":"Measure (data warehouse)","level":2,"score":0.3506999909877777},{"id":"https://openalex.org/C163175372","wikidata":"https://www.wikidata.org/wiki/Q3339222","display_name":"Linear model","level":2,"score":0.3425000011920929},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.319599986076355},{"id":"https://openalex.org/C168167062","wikidata":"https://www.wikidata.org/wiki/Q1117970","display_name":"Component (thermodynamics)","level":2,"score":0.3082999885082245},{"id":"https://openalex.org/C162392398","wikidata":"https://www.wikidata.org/wiki/Q272404","display_name":"Finite set","level":2,"score":0.3043000102043152},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.3037000000476837},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.29760000109672546},{"id":"https://openalex.org/C19539793","wikidata":"https://www.wikidata.org/wiki/Q7449609","display_name":"Semiparametric regression","level":3,"score":0.2863999903202057},{"id":"https://openalex.org/C2776502983","wikidata":"https://www.wikidata.org/wiki/Q690182","display_name":"Contrast (vision)","level":2,"score":0.2770000100135803},{"id":"https://openalex.org/C122383733","wikidata":"https://www.wikidata.org/wiki/Q865920","display_name":"Approximation error","level":2,"score":0.2766999900341034},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.27059999108314514},{"id":"https://openalex.org/C14646407","wikidata":"https://www.wikidata.org/wiki/Q1430750","display_name":"Bellman equation","level":2,"score":0.27000001072883606},{"id":"https://openalex.org/C41045048","wikidata":"https://www.wikidata.org/wiki/Q202843","display_name":"Linear programming","level":2,"score":0.26429998874664307},{"id":"https://openalex.org/C77553402","wikidata":"https://www.wikidata.org/wiki/Q13222579","display_name":"Upper and lower bounds","level":2,"score":0.25850000977516174},{"id":"https://openalex.org/C102634674","wikidata":"https://www.wikidata.org/wiki/Q868473","display_name":"Smoothness","level":2,"score":0.25459998846054077},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.25290000438690186}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:arXiv.org:2609.37209","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.37209","pdf_url":"https://arxiv.org/pdf/2609.37209","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.2609.37209","doi":"https://doi.org/10.48550/arxiv.2609.37209","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2609.37209","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":"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"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2609.37209","doi":null,"is_oa":true,"landing_page_url":"https://arxiv.org/abs/2609.37209","pdf_url":"https://arxiv.org/pdf/2609.37209","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":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W7214928710.pdf","grobid_xml":"https://content.openalex.org/works/W7214928710.grobid-xml"},"referenced_works_count":42,"referenced_works":["https://openalex.org/W135795571","https://openalex.org/W1513786872","https://openalex.org/W1538452572","https://openalex.org/W1614597761","https://openalex.org/W1778702646","https://openalex.org/W1938444229","https://openalex.org/W1973286131","https://openalex.org/W1976576119","https://openalex.org/W1978810924","https://openalex.org/W2012547817","https://openalex.org/W2013784666","https://openalex.org/W2022392475","https://openalex.org/W2030189596","https://openalex.org/W2051434435","https://openalex.org/W2052044664","https://openalex.org/W2054434031","https://openalex.org/W2063368770","https://openalex.org/W2064868327","https://openalex.org/W2075621176","https://openalex.org/W2076541756","https://openalex.org/W2078658856","https://openalex.org/W2080872718","https://openalex.org/W2084572036","https://openalex.org/W2111396966","https://openalex.org/W2113963519","https://openalex.org/W2117594039","https://openalex.org/W2117896839","https://openalex.org/W2318460400","https://openalex.org/W2914527946","https://openalex.org/W2990830025","https://openalex.org/W3089065092","https://openalex.org/W3123436326","https://openalex.org/W3138939485","https://openalex.org/W3195140297","https://openalex.org/W3210904070","https://openalex.org/W4382518033","https://openalex.org/W4391631327","https://openalex.org/W4403796339","https://openalex.org/W4404781622","https://openalex.org/W4404792842","https://openalex.org/W7163004758","https://openalex.org/W7166879153"],"related_works":["https://openalex.org/W4417088305","https://openalex.org/W2809697779","https://openalex.org/W4414809479","https://openalex.org/W6885057546","https://openalex.org/W7202378854","https://openalex.org/W4313050493","https://openalex.org/W4400022770","https://openalex.org/W1578240932","https://openalex.org/W7096606896","https://openalex.org/W3010157058"],"abstract_inverted_index":{"Pairwise":[0],"losses":[1,23],"are":[2,12],"increasingly":[3],"used":[4],"for":[5,113],"reward":[6,44,64,84],"learning":[7,45],"even":[8],"when":[9,180],"pointwise":[10,25],"rewards":[11,55,179],"observed,":[13],"with":[14,57,196],"mixed":[15],"empirical":[16],"results.":[17],"When":[18],"and":[19,93,115,119,137],"why":[20],"do":[21],"pairwise":[22],"outperform":[24],"losses?":[26],"We":[27,47,76],"study":[28],"this":[29],"question":[30],"in":[31,133],"a":[32,67,78,89,104,139,150,188],"grouped":[33],"offline":[34],"contextual-bandit":[35],"setting":[36],"allowing":[37],"multiple":[38],"actions":[39,70,146],"per":[40],"context,":[41,149],"capturing":[42,100],"many":[43],"scenarios.":[46],"compare":[48],"Value":[49,58],"Regression":[50,60],"(VR),":[51],"which":[52,62,193],"regresses":[53,63],"observed":[54],"pointwise,":[56],"Difference":[59],"(VDR),":[61],"differences":[65],"between":[66],"pair":[68],"of":[69,88],"sampled":[71],"under":[72],"the":[73,82,86,130,134,154,181],"same":[74],"context.":[75],"consider":[77],"semiparametric":[79],"model":[80],"where":[81],"mean":[83],"is":[85,183],"sum":[87],"learnable":[90],"action-dependent":[91],"component":[92],"an":[94],"arbitrary":[95],"context-dependent":[96],"yet":[97],"action-independent":[98],"nuisance,":[99],"context-specific":[101],"disturbances.":[102],"Using":[103],"unified":[105],"localized":[106],"analysis,":[107],"we":[108,194],"prove":[109],"finite-sample":[110],"regression":[111],"guarantees":[112],"finite":[114,126],"linear":[116,159],"function":[117],"classes":[118],"translate":[120],"them":[121],"into":[122],"offline-regret":[123],"bounds.":[124],"For":[125,158],"classes,":[127,160],"VDR":[128],"eliminates":[129],"misspecification":[131,182],"term":[132,142],"VR":[135,156],"bound":[136],"improves":[138],"reward-scale-dependent":[140],"error":[141],"by":[143],"averaging":[144],"over":[145],"within":[147],"each":[148],"benefit":[151],"absent":[152],"from":[153],"corresponding":[155],"term.":[157],"neither":[161],"method":[162],"uniformly":[163],"dominates:":[164],"within-context":[165],"differencing":[166],"removes":[167],"nuisance-induced":[168],"bias":[169],"but":[170],"may":[171],"increase":[172],"estimation":[173],"variance":[174],"relative":[175],"to":[176],"using":[177],"absolute":[178],"sufficiently":[184],"low.":[185],"This":[186],"yields":[187],"feature":[189],"geometry-dependent":[190],"bias-variance":[191],"tradeoff,":[192],"corroborate":[195],"numerical":[197],"experiments.":[198]},"counts_by_year":[],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2026-10-01T00:00:00"}