{"id":"https://openalex.org/W2294845706","doi":null,"title":"Policy Improvement Methods: Between Black-Box Optimization and Episodic Reinforcement Learning","display_name":"Policy Improvement Methods: Between Black-Box Optimization and Episodic Reinforcement Learning","publication_year":2012,"publication_date":"2012-10-04","ids":{"openalex":"https://openalex.org/W2294845706","mag":"2294845706"},"language":"en","primary_location":{"id":"pmh:oai:CiteSeerX.psu:10.1.1.393.981","doi":null,"is_oa":false,"landing_page_url":"http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.393.981","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"http://hal.inria.fr/docs/00/73/84/63/PDF/rl_blackbox_new_HAL.pdf","raw_type":"text"},"volume":null,"type":"article","is_traditional":true,"indexed_in":[],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://hal.science/hal-00738463","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5017689065","display_name":"Freek Stulp","orcid":"https://orcid.org/0000-0001-9555-9517","observed_orcids":["https://orcid.org/0000-0001-9555-9517"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Freek Stulp","raw_affiliation_strings":["Robotique et Vision"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Robotique et Vision","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5042850624","display_name":"Olivier Sigaud","orcid":"https://orcid.org/0000-0002-8544-0229","observed_orcids":["https://orcid.org/0000-0002-8544-0229"]},"institutions":[{"id":"https://openalex.org/I4210150358","display_name":"Institut Syst\u00e8mes Intelligents et de Robotique","ror":"https://ror.org/05neq8668","country_code":"FR","type":"facility","lineage":["https://openalex.org/I1294671590","https://openalex.org/I154526488","https://openalex.org/I39804081","https://openalex.org/I4210150358","https://openalex.org/I4210159245"]}],"countries":["FR"],"is_corresponding":false,"raw_author_name":"Olivier Sigaud","raw_affiliation_strings":["Institut des Syst\u00e8mes Intelligents et de Robotique"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Institut des Syst\u00e8mes Intelligents et de Robotique","institution_ids":["https://openalex.org/I4210150358"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":2.9019,"has_fulltext":false,"cited_by_count":66,"citation_normalized_percentile":{"value":0.91482169,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":93,"max":99},"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/T10462","display_name":"Reinforcement Learning in Robotics","score":0.9994999766349792,"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":0.9994999766349792,"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/T12794","display_name":"Adaptive Dynamic Programming Control","score":9.999999747378752e-05,"subfield":{"id":"https://openalex.org/subfields/2207","display_name":"Control and Systems Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11195","display_name":"Simulation Techniques and Applications","score":0.0,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/policy-improvement","display_name":"policy improvement","score":0.859000027179718},{"id":"https://openalex.org/keywords/black-box-optimization","display_name":"black-box optimization","score":0.9890000224113464},{"id":"https://openalex.org/keywords/episodic-reinforcement-learning","display_name":"episodic reinforcement learning","score":0.8970000147819519},{"id":"https://openalex.org/keywords/cma-es","display_name":"CMA-ES","score":0.8240000009536743},{"id":"https://openalex.org/keywords/parameter-perturbation","display_name":"parameter perturbation","score":0.7979999780654907},{"id":"https://openalex.org/keywords/gradient-estimation","display_name":"gradient estimation","score":0.7710000276565552},{"id":"https://openalex.org/keywords/direct-policy-search","display_name":"direct policy search","score":0.7820000052452087},{"id":"https://openalex.org/keywords/actor-critic","display_name":"actor-critic","score":0.9139999747276306},{"id":"https://openalex.org/keywords/algorithmic-properties","display_name":"algorithmic properties","score":0.49900001287460327},{"id":"https://openalex.org/keywords/convergence-rate","display_name":"convergence rate","score":0.6240000128746033}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.7870957851409912},{"id":"https://openalex.org/C94966114","wikidata":"https://www.wikidata.org/wiki/Q29256","display_name":"Black box","level":2,"score":0.6864012479782104},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.495837539434433},{"id":"https://openalex.org/C67203356","wikidata":"https://www.wikidata.org/wiki/Q1321905","display_name":"Reinforcement","level":2,"score":0.49413326382637024},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4020160734653473},{"id":"https://openalex.org/C15744967","wikidata":"https://www.wikidata.org/wiki/Q9418","display_name":"Psychology","level":0,"score":0.270153284072876},{"id":"https://openalex.org/C77805123","wikidata":"https://www.wikidata.org/wiki/Q161272","display_name":"Social psychology","level":1,"score":0.16697818040847778}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:CiteSeerX.psu:10.1.1.393.981","doi":null,"is_oa":false,"landing_page_url":"http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.393.981","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"http://hal.inria.fr/docs/00/73/84/63/PDF/rl_blackbox_new_HAL.pdf","raw_type":"text"},{"id":"pmh:oai:HAL:hal-00738463v1","doi":null,"is_oa":true,"landing_page_url":"https://hal.science/hal-00738463","pdf_url":null,"source":{"id":"https://openalex.org/S4306402512","display_name":"HAL","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I4405252976","host_organization_name":"Centre pour la Communication Scientifique Directe","host_organization_lineage":["https://openalex.org/I4405252976"],"host_organization_lineage_names":[],"type":"repository"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"2012","raw_type":"info:eu-repo/semantics/preprint"}],"best_oa_location":{"id":"pmh:oai:HAL:hal-00738463v1","doi":null,"is_oa":true,"landing_page_url":"https://hal.science/hal-00738463","pdf_url":null,"source":{"id":"https://openalex.org/S4306402512","display_name":"HAL","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I4405252976","host_organization_name":"Centre pour la Communication Scientifique Directe","host_organization_lineage":["https://openalex.org/I4405252976"],"host_organization_lineage_names":[],"type":"repository"},"license":"other-oa","license_id":"https://openalex.org/licenses/other-oa","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"2012","raw_type":"info:eu-repo/semantics/preprint"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":34,"referenced_works":["https://openalex.org/W155952036","https://openalex.org/W1262227701","https://openalex.org/W1480347379","https://openalex.org/W1503196123","https://openalex.org/W1533535072","https://openalex.org/W1586251222","https://openalex.org/W1601081659","https://openalex.org/W1601974704","https://openalex.org/W1859403577","https://openalex.org/W1914583973","https://openalex.org/W1925816294","https://openalex.org/W2012392077","https://openalex.org/W2026659355","https://openalex.org/W2060914855","https://openalex.org/W2098524868","https://openalex.org/W2108051346","https://openalex.org/W2108474901","https://openalex.org/W2109572794","https://openalex.org/W2110304639","https://openalex.org/W2112036188","https://openalex.org/W2116339921","https://openalex.org/W2118459920","https://openalex.org/W2119717200","https://openalex.org/W2121863487","https://openalex.org/W2123327324","https://openalex.org/W2125612430","https://openalex.org/W2133932631","https://openalex.org/W2142916680","https://openalex.org/W2146737184","https://openalex.org/W2146957157","https://openalex.org/W2148067905","https://openalex.org/W2150304264","https://openalex.org/W2154549708","https://openalex.org/W2172968643"],"related_works":["https://openalex.org/W4298008065","https://openalex.org/W2085627234","https://openalex.org/W3112829623","https://openalex.org/W7100872271","https://openalex.org/W2397513399","https://openalex.org/W7148303674","https://openalex.org/W2142383762","https://openalex.org/W2098524868","https://openalex.org/W2945269249","https://openalex.org/W4401647482"],"abstract_inverted_index":{"Policy":[0],"improvement":[1,64],"methods":[2,39,184],"seek":[3],"to":[4,13,22,42,49,58,84,103,188,199,281,319],"optimize":[5],"the":[6,60,117,151,154,200,233,246,303,311],"parameters":[7],"of":[8,62,93,106,132,137,153,156,175,232,257,290,310,314],"a":[9,14,162,220,229,283],"policy":[10,51,63,145],"with":[11,204],"respect":[12],"utility":[15],"function.":[16],"There":[17],"are":[18,36,55,294],"two":[19,197],"main":[20,312],"approaches":[21,350],"performing":[23],"this":[24,97,104,176,193,280,315],"optimization:":[25],"reinforcement":[26],"learning":[27],"(RL)":[28],"and":[29,73,78,121,139,169,186,260,286,333,338,342,348],"black-box":[30],"optimization":[31,38],"(BBO).":[32],"Whereas":[33],"BBO":[34,74,221,273,306,347],"algorithms":[35,54,160,165],"generic":[37],"that,":[40,268],"due":[41],"there":[43,79],"generality,":[44],"may":[45,328],"also":[46],"be":[47,282,329],"applied":[48],"optimizing":[50],"parameters,":[52],"RL":[53,72,120,183,349],"specifically":[56],"tailored":[57],"leveraging":[59],"structure":[61],"problems.":[65],"In":[66,96],"recent":[67],"years,":[68],"benchmark":[69],"comparisons":[70],"between":[71,119],"have":[75],"been":[76,81],"made,":[77,330],"has":[80,181],"several":[82,101],"attempts":[83],"specify":[85],"which":[86,91,208,325],"approach":[87],"works":[88],"best":[89],"for":[90,269,297,305],"types":[92],"problem":[94],"classes.":[95],"article,":[98],"we":[99,212,276],"make":[100],"contributions":[102,313],"line":[105],"research:":[107],"1)":[108],"We":[109,148,191,215],"define":[110],"four":[111],"algorithmic":[112,322],"properties":[113],"that":[114,179,217,226,245],"further":[115],"clarify":[116],"relationship":[118],"BBO:":[122],"action-perturbation":[123],"vs.":[124,128,135,143],"parameter-perturbation,":[125],"gradient":[126],"estimation":[127],"rewardweighted":[129],"averaging,":[130],"use":[131,136,335],"only":[133,344],"rewards":[134,138],"state":[140],"information,":[141],"actor-critic":[142],"direct":[144],"search.":[146],"2)":[147],"show":[149,216],"how":[150],"chronology":[152],"derivation":[155],"ever":[157],"more":[158,224],"powerful":[159],"displays":[161],"trend":[163,177,194],"towards":[164],"based":[166],"on":[167,251],"parameter-perturbation":[168],"reward-weighted":[170],"averaging.":[171],"A":[172],"striking":[173],"feature":[174],"is":[178,219,228,317],"it":[180,227],"moved":[182],"closer":[185,187],"BBO.":[189,298],"3)":[190],"continue":[192],"by":[195],"applying":[196],"modifications":[198],"state-of-the-art":[201],"\"Policy":[202],"Improvement":[203],"Path":[205],"Integrals\"":[206],"(PI2),":[207],"yields":[209],"an":[210,288,321],"algorithm":[211],"denote":[213],"PIBB.":[214],"PIBB":[218,248,332],"algorithm,":[222],"and,":[223],"specifically,":[225],"special":[230],"case":[231,304],"\"Covariance":[234],"Matrix":[235],"Adaptation":[236],"-":[237],"Evolutionary":[238],"Strategy\"":[239],"algorithm.":[240],"Our":[241],"empirical":[242],"evaluation":[243,253,266],"demonstrates":[244],"simpler":[247],"outperforms":[249,274],"PI2":[250,334],"simple":[252],"tasks":[254,293],"in":[255,324,345],"terms":[256],"convergence":[258],"speed":[259],"final":[261],"cost.":[262],"4)":[263],"Although":[264],"our":[265],"implies":[267],"these":[270,292],"five":[271],"tasks,":[272],"RL,":[275,308],"do":[277],"not":[278],"hold":[279],"general":[284],"statement,":[285],"provide":[287,320],"analysis":[289],"why":[291],"particularly":[295],"well-suited":[296],"Thus,":[299],"rather":[300,318],"than":[301],"making":[302],"or":[307],"one":[309],"article":[316],"framework":[323],"such":[326],"cases":[327],"as":[331],"identical":[336],"perturbation":[337],"parameter":[339],"update":[340],"methods,":[341],"differ":[343],"being":[346],"respectively.":[351]},"counts_by_year":[{"year":2025,"cited_by_count":2},{"year":2024,"cited_by_count":5},{"year":2023,"cited_by_count":2},{"year":2022,"cited_by_count":4},{"year":2021,"cited_by_count":4},{"year":2020,"cited_by_count":9},{"year":2019,"cited_by_count":5},{"year":2018,"cited_by_count":9},{"year":2017,"cited_by_count":11},{"year":2016,"cited_by_count":5},{"year":2015,"cited_by_count":2},{"year":2014,"cited_by_count":6},{"year":2013,"cited_by_count":2}],"updated_date":"2026-11-08T12:12:28.789561","created_date":"2016-06-24T00:00:00"}