{"id":"https://openalex.org/W7140282989","doi":"https://doi.org/10.1016/j.artint.2026.104525","title":"Partial identifiability and misspecification in inverse reinforcement learning","display_name":"Partial identifiability and misspecification in inverse reinforcement learning","publication_year":2026,"publication_date":"2026-03-25","ids":{"openalex":"https://openalex.org/W7140282989","doi":"https://doi.org/10.1016/j.artint.2026.104525"},"language":"en","primary_location":{"id":"doi:10.1016/j.artint.2026.104525","is_oa":true,"landing_page_url":"https://doi.org/10.1016/j.artint.2026.104525","pdf_url":null,"source":{"id":"https://openalex.org/S196139623","display_name":"Artificial Intelligence","issn_l":"0004-3702","issn":["0004-3702","1872-7921"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320990","host_organization_name":"Elsevier BV","host_organization_lineage":["https://openalex.org/P4310320990"],"host_organization_lineage_names":["Elsevier BV"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Artificial Intelligence","raw_type":"journal-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://doi.org/10.1016/j.artint.2026.104525","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Joar Skalse","orcid":"https://orcid.org/0009-0005-0848-7188"},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]},{"id":"https://openalex.org/I4210146410","display_name":"Science Oxford","ror":"https://ror.org/04j8yhy50","country_code":"GB","type":"nonprofit","lineage":["https://openalex.org/I4210146410"]}],"countries":["GB"],"is_corresponding":true,"raw_author_name":"Joar Skalse","raw_affiliation_strings":["Department of Computer Science, University of Oxford, The United Kingdom"],"raw_orcid":"https://orcid.org/0009-0005-0848-7188","affiliations":[{"raw_affiliation_string":"Department of Computer Science, University of Oxford, The United Kingdom","institution_ids":["https://openalex.org/I4210146410","https://openalex.org/I40120149"]}]},{"author_position":"last","author":{"id":null,"display_name":"Alessandro Abate","orcid":"https://orcid.org/0000-0002-5627-9093"},"institutions":[{"id":"https://openalex.org/I40120149","display_name":"University of Oxford","ror":"https://ror.org/052gg0110","country_code":"GB","type":"education","lineage":["https://openalex.org/I40120149"]},{"id":"https://openalex.org/I4210146410","display_name":"Science Oxford","ror":"https://ror.org/04j8yhy50","country_code":"GB","type":"nonprofit","lineage":["https://openalex.org/I4210146410"]}],"countries":["GB"],"is_corresponding":false,"raw_author_name":"Alessandro Abate","raw_affiliation_strings":["Department of Computer Science, University of Oxford, The United Kingdom"],"raw_orcid":"https://orcid.org/0000-0002-5627-9093","affiliations":[{"raw_affiliation_string":"Department of Computer Science, University of Oxford, The United Kingdom","institution_ids":["https://openalex.org/I4210146410","https://openalex.org/I40120149"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I40120149","https://openalex.org/I4210146410"],"apc_list":{"value":3670,"currency":"USD","value_usd":3670},"apc_paid":{"value":3670,"currency":"USD","value_usd":3670},"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.63484913,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"356","issue":null,"first_page":"104525","last_page":"104525"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10462","display_name":"Reinforcement Learning in Robotics","score":0.6407999992370605,"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.6407999992370605,"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/T12101","display_name":"Advanced Bandit Algorithms Research","score":0.09809999912977219,"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/T10042","display_name":"Neural and Behavioral Psychology Studies","score":0.031700000166893005,"subfield":{"id":"https://openalex.org/subfields/2805","display_name":"Cognitive Neuroscience"},"field":{"id":"https://openalex.org/fields/28","display_name":"Neuroscience"},"domain":{"id":"https://openalex.org/domains/1","display_name":"Life Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/identifiability","display_name":"Identifiability","score":0.9688000082969666},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.6495000123977661},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"Reinforcement learning","score":0.5295000076293945},{"id":"https://openalex.org/keywords/class","display_name":"Class (philosophy)","score":0.49459999799728394},{"id":"https://openalex.org/keywords/inverse","display_name":"Inverse","score":0.40119999647140503},{"id":"https://openalex.org/keywords/inverse-problem","display_name":"Inverse problem","score":0.3107999861240387}],"concepts":[{"id":"https://openalex.org/C122770356","wikidata":"https://www.wikidata.org/wiki/Q1656753","display_name":"Identifiability","level":2,"score":0.9688000082969666},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.6495000123977661},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.5295000076293945},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.49459999799728394},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4465999901294708},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.43220001459121704},{"id":"https://openalex.org/C207467116","wikidata":"https://www.wikidata.org/wiki/Q4385666","display_name":"Inverse","level":2,"score":0.40119999647140503},{"id":"https://openalex.org/C149782125","wikidata":"https://www.wikidata.org/wiki/Q160039","display_name":"Econometrics","level":1,"score":0.3968000113964081},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3427000045776367},{"id":"https://openalex.org/C135252773","wikidata":"https://www.wikidata.org/wiki/Q1567213","display_name":"Inverse problem","level":2,"score":0.3107999861240387},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.30309998989105225},{"id":"https://openalex.org/C2778572836","wikidata":"https://www.wikidata.org/wiki/Q380933","display_name":"Space (punctuation)","level":2,"score":0.3001999855041504},{"id":"https://openalex.org/C148043351","wikidata":"https://www.wikidata.org/wiki/Q4456944","display_name":"Current (fluid)","level":2,"score":0.27869999408721924},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.27390000224113464},{"id":"https://openalex.org/C151201525","wikidata":"https://www.wikidata.org/wiki/Q177239","display_name":"Limit (mathematics)","level":2,"score":0.26969999074935913},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.25540000200271606}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1016/j.artint.2026.104525","is_oa":true,"landing_page_url":"https://doi.org/10.1016/j.artint.2026.104525","pdf_url":null,"source":{"id":"https://openalex.org/S196139623","display_name":"Artificial Intelligence","issn_l":"0004-3702","issn":["0004-3702","1872-7921"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320990","host_organization_name":"Elsevier BV","host_organization_lineage":["https://openalex.org/P4310320990"],"host_organization_lineage_names":["Elsevier BV"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Artificial Intelligence","raw_type":"journal-article"}],"best_oa_location":{"id":"doi:10.1016/j.artint.2026.104525","is_oa":true,"landing_page_url":"https://doi.org/10.1016/j.artint.2026.104525","pdf_url":null,"source":{"id":"https://openalex.org/S196139623","display_name":"Artificial Intelligence","issn_l":"0004-3702","issn":["0004-3702","1872-7921"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320990","host_organization_name":"Elsevier BV","host_organization_lineage":["https://openalex.org/P4310320990"],"host_organization_lineage_names":["Elsevier BV"],"type":"journal"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Artificial Intelligence","raw_type":"journal-article"},"sustainable_development_goals":[{"display_name":"Peace, Justice and strong institutions","score":0.539644718170166,"id":"https://metadata.un.org/sdg/16"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":17,"referenced_works":["https://openalex.org/W2004717177","https://openalex.org/W2027047704","https://openalex.org/W2084762552","https://openalex.org/W2093002938","https://openalex.org/W2134491302","https://openalex.org/W2205847592","https://openalex.org/W2333311386","https://openalex.org/W2604382266","https://openalex.org/W2949206666","https://openalex.org/W2963762522","https://openalex.org/W3011865677","https://openalex.org/W3138984732","https://openalex.org/W3157893055","https://openalex.org/W4210909794","https://openalex.org/W4246892532","https://openalex.org/W4382318927","https://openalex.org/W4385569890"],"related_works":[],"abstract_inverted_index":{"\u2022":[0,43,86,123,155,173,217],"Research":[1,44,87,124,156,174,218],"highlight":[2,45,88,125,157,175,219],"1:":[3],"We":[4,47,90,127,159,177,221,440],"present":[5],"formal":[6,498],"frameworks":[7,93],"for":[8,282,425,486],"reasoning":[9,487],"about":[10,471,488],"partial":[11,34,97,131,407,489,508],"identifiability":[12,35,98,132,408,490,509],"and":[13,19,36,68,75,99,120,312,353,359,409,417,444,491,510],"misspecification":[14,37,100,209,410,492,511],"in":[15,109,116,139,198,202,242,246,323,375,411,435,493],"inverse":[16,112],"reinforcement":[17,113],"learning,":[18],"use":[20,91],"these":[21,62,78],"to":[22,31,83,94,168,206,251,267,325,343,362,389,394,468,479,504],"derive":[23,32,48,506],"several":[24,283,497],"mathematical":[25,404],"tools":[26,499],"which":[27,294,381],"can":[28,248,501],"be":[29,249,378,502],"used":[30,503],"the":[33,54,96,110,130,134,140,150,191,199,211,235,243,256,304,347,372,383,419,422,428,436,451,460,472,507],"robustness":[38,101,512],"of":[39,51,56,102,118,133,142,171,183,210,227,255,261,286,320,339,406,421,427,459,513,521],"new":[40,514],"behavioural":[41,104,137,163,184,192,228,236,337,373,429,462],"models.":[42],"2:":[46],"a":[49,69,180,224,269,274,298,316,336,366,402,483],"class":[50,182,226],"pseudometrics":[52,63,79],"on":[53,72],"space":[55],"all":[57,103,426],"reward":[58,270,292,305,423,473,522],"functions,":[59],"show":[60,76,128,160,178,222],"that":[61,77,106,129,149,161,179,194,223,238,303,313,371,385,431,447,465,500],"induce":[64],"both":[65],"an":[66,331],"upper":[67],"lower":[70],"bound":[71],"worst-case":[73],"regret,":[74],"are":[80,107,165,195,204,239,289,295,432],"unique":[81],"(up":[82],"bilipschitz":[84],"equivalence).":[85],"3:":[89],"our":[92],"characterise":[95,416],"models":[105,138,164,185,193,229,237,430,463],"popular":[108],"current":[111,200,244,437],"learning":[114,523],"literature":[115],"terms":[117],"necessary":[119,443],"sufficient":[121,445],"conditions.":[122],"4:":[126],"most":[135,196,240,433],"common":[136,197,241,434],"limit":[141],"infinite":[143],"data":[144],"is":[145,152,266,280,307,356],"mostly":[146],"unproblematic,":[147],"provided":[148],"model":[151,338,374,466],"not":[153,188,232],"misspecified.":[154],"5:":[158],"many":[162],"highly":[166],"robust":[167],"certain":[169,317],"forms":[170],"misspecification.":[172],"6:":[176],"broad":[181,225],"(including,":[186,230],"but":[187,231],"limited":[189,233],"to,":[190,234],"literature)":[201,245],"principle":[203,247],"sensitive":[205,250],"arbitrarily":[207,252],"small":[208,253],"discount":[212],"factor":[213],"or":[214,517],"transition":[215],"function.":[216],"7:":[220],"perturbations":[254],"observed":[257,452],"policy.":[258],"The":[259],"aim":[260],"Inverse":[262],"Reinforcement":[263],"Learning":[264],"(IRL)":[265],"infer":[268,326],"function":[271,306,424,474],"R":[272,327,344,475],"from":[273,328,457],"policy":[275,454],"\u03c0":[276,329,341],".":[277,345,476],"This":[278,369],"problem":[279],"difficult,":[281],"reasons.":[284],"First":[285],"all,":[287],"there":[288],"typically":[290],"multiple":[291],"functions":[293],"compatible":[296],"with":[297,365,496],"given":[299],"policy;":[300],"this":[301,398],"means":[302,370],"only":[308],"partially":[309],"identifiable":[310],",":[311,330,380],"IRL":[314,332,438,515],"contains":[315],"fundamental":[318],"degree":[319],"ambiguity.":[321],"Secondly,":[322],"order":[324],"algorithm":[333],"must":[334],"have":[335],"how":[340,450],"relates":[342],"However,":[346],"true":[348],"relationship":[349],"between":[350],"human":[351,354],"preferences":[352],"behaviour":[355],"very":[357],"complex,":[358],"practically":[360],"impossible":[361],"fully":[363,415],"capture":[364],"simple":[367],"model.":[368],"practice":[376],"will":[377],"misspecified":[379],"raises":[382],"worry":[384],"it":[386],"might":[387],"lead":[388],"unsound":[390],"inferences":[391,470],"if":[392],"applied":[393],"real-world":[395],"data.":[396],"In":[397,477],"paper,":[399],"we":[400,414,481],"provide":[401,442],"comprehensive":[403],"analysis":[405],"IRL.":[412],"Specifically,":[413],"quantify":[418],"ambiguity":[420],"literature.":[439],"also":[441],"conditions":[446],"describe":[448],"precisely":[449],"demonstrator":[453],"may":[455],"differ":[456],"each":[458],"standard":[461],"before":[464],"leads":[467],"faulty":[469],"addition":[478],"this,":[480],"introduce":[482],"cohesive":[484],"framework":[485],"IRL,":[494],"together":[495],"easily":[505],"models,":[516],"analyse":[518],"other":[519],"kinds":[520],"algorithms.":[524]},"counts_by_year":[],"updated_date":"2026-04-25T06:01:18.069262","created_date":"2025-10-10T00:00:00"}