{"id":"https://openalex.org/W7225830865","doi":"https://doi.org/10.48550/arxiv.2611.13480","title":"Sharp Sample Complexity of KL-Regularized Reinforcement Learning","display_name":"Sharp Sample Complexity of KL-Regularized Reinforcement Learning","publication_year":2026,"publication_date":"2026-11-10","ids":{"openalex":"https://openalex.org/W7225830865","doi":"https://doi.org/10.48550/arxiv.2611.13480"},"language":"en","primary_location":{"id":"doi:10.48550/arxiv.2611.13480","doi":"https://doi.org/10.48550/arxiv.2611.13480","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.13480","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"volume":null,"type":"preprint","is_traditional":true,"indexed_in":["datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://doi.org/10.48550/arxiv.2611.13480","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5080442924","display_name":"Andrew Liu","orcid":"https://orcid.org/0000-0003-2032-9842","observed_orcids":["https://orcid.org/0000-0003-2032-9842"]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Liu, Andrew","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5080442924"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"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/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/T12101","display_name":"Advanced Bandit Algorithms Research","score":0.00019999999494757503,"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/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":[],"concepts":[{"id":"https://openalex.org/C149728462","wikidata":"https://www.wikidata.org/wiki/Q751319","display_name":"Minimax","level":2,"score":0.7307000160217285},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.6923999786376953},{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.6442000269889832},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.5461000204086304},{"id":"https://openalex.org/C207390915","wikidata":"https://www.wikidata.org/wiki/Q1230525","display_name":"Divergence (linguistics)","level":2,"score":0.5383999943733215},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.489300012588501},{"id":"https://openalex.org/C77553402","wikidata":"https://www.wikidata.org/wiki/Q13222579","display_name":"Upper and lower bounds","level":2,"score":0.4860999882221222},{"id":"https://openalex.org/C2777027219","wikidata":"https://www.wikidata.org/wiki/Q1284190","display_name":"Constant (computer programming)","level":2,"score":0.453900009393692},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.40869998931884766},{"id":"https://openalex.org/C198531522","wikidata":"https://www.wikidata.org/wiki/Q485146","display_name":"Sample (material)","level":2,"score":0.4066999852657318},{"id":"https://openalex.org/C28761237","wikidata":"https://www.wikidata.org/wiki/Q7805321","display_name":"Time horizon","level":2,"score":0.3935000002384186},{"id":"https://openalex.org/C2778755073","wikidata":"https://www.wikidata.org/wiki/Q10858537","display_name":"Scale (ratio)","level":2,"score":0.37779998779296875},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.37529999017715454},{"id":"https://openalex.org/C196083921","wikidata":"https://www.wikidata.org/wiki/Q7915758","display_name":"Variance (accounting)","level":2,"score":0.3725000023841858},{"id":"https://openalex.org/C181789720","wikidata":"https://www.wikidata.org/wiki/Q4812191","display_name":"Asymptotically optimal algorithm","level":2,"score":0.36880001425743103},{"id":"https://openalex.org/C133939421","wikidata":"https://www.wikidata.org/wiki/Q6865379","display_name":"Minimax estimator","level":4,"score":0.3650999963283539},{"id":"https://openalex.org/C129848803","wikidata":"https://www.wikidata.org/wiki/Q2564360","display_name":"Sample size determination","level":2,"score":0.36489999294281006},{"id":"https://openalex.org/C2780586882","wikidata":"https://www.wikidata.org/wiki/Q7520643","display_name":"Simple (philosophy)","level":2,"score":0.36230000853538513},{"id":"https://openalex.org/C2780813799","wikidata":"https://www.wikidata.org/wiki/Q3274237","display_name":"Zero (linguistics)","level":2,"score":0.36000001430511475},{"id":"https://openalex.org/C2777735758","wikidata":"https://www.wikidata.org/wiki/Q817765","display_name":"Path (computing)","level":2,"score":0.3336000144481659},{"id":"https://openalex.org/C65778772","wikidata":"https://www.wikidata.org/wiki/Q12345341","display_name":"Asymptotic distribution","level":3,"score":0.3310999870300293},{"id":"https://openalex.org/C159176650","wikidata":"https://www.wikidata.org/wiki/Q43261","display_name":"Horizon","level":2,"score":0.3102000057697296},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.29089999198913574},{"id":"https://openalex.org/C17648541","wikidata":"https://www.wikidata.org/wiki/Q2265984","display_name":"Efficiency","level":3,"score":0.2768000066280365},{"id":"https://openalex.org/C91575142","wikidata":"https://www.wikidata.org/wiki/Q1971426","display_name":"Optimal control","level":2,"score":0.2750999927520752},{"id":"https://openalex.org/C171752962","wikidata":"https://www.wikidata.org/wiki/Q255166","display_name":"Kullback\u2013Leibler divergence","level":2,"score":0.2678000032901764},{"id":"https://openalex.org/C165646398","wikidata":"https://www.wikidata.org/wiki/Q3755281","display_name":"Minimum-variance unbiased estimator","level":3,"score":0.2646999955177307},{"id":"https://openalex.org/C48103436","wikidata":"https://www.wikidata.org/wiki/Q599031","display_name":"State (computer science)","level":2,"score":0.26420000195503235},{"id":"https://openalex.org/C2777212361","wikidata":"https://www.wikidata.org/wiki/Q5127848","display_name":"Class (philosophy)","level":2,"score":0.26260000467300415},{"id":"https://openalex.org/C140779682","wikidata":"https://www.wikidata.org/wiki/Q210868","display_name":"Sampling (signal processing)","level":3,"score":0.2558000087738037},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.2524999976158142},{"id":"https://openalex.org/C971699","wikidata":"https://www.wikidata.org/wiki/Q1132714","display_name":"Delta method","level":3,"score":0.2500999867916107}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.48550/arxiv.2611.13480","doi":"https://doi.org/10.48550/arxiv.2611.13480","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.13480","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"doi:10.48550/arxiv.2611.13480","doi":"https://doi.org/10.48550/arxiv.2611.13480","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2611.13480","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":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":[],"abstract_inverted_index":{"KL-regularized":[0],"reinforcement":[1],"learning,":[2],"the":[3,17,24,34,74,86,103,149,159,164,171,192,207,210],"objective":[4],"behind":[5],"maximum-entropy":[6],"RL":[7,9],"and":[8,33,78,121,189,209],"from":[10,112],"human":[11],"feedback,":[12],"is":[13,28,39,50,110,144,177,197],"variational":[14],"inference":[15],"in":[16,73],"graphical":[18],"model":[19],"of":[20,36,163,170,194],"\"control":[21],"as":[22,153],"inference\":":[23],"regularized":[25],"optimal":[26,198],"policy":[27,38,108],"a":[29,37,42,64,96,107,145,167],"posterior":[30],"over":[31],"trajectories,":[32],"suboptimality":[35],"$\u03b2$":[40],"times":[41],"KL":[43],"divergence":[44],"between":[45],"path":[46],"measures.":[47],"However,":[48],"it":[49,56],"not":[51],"known":[52],"how":[53],"many":[54],"samples":[55,195],"takes":[57],"to":[58,130],"learn":[59],"this":[60,90,143],"posterior,":[61],"even":[62,125],"with":[63,71,99,116,134,148],"generative":[65],"model.":[66],"Earlier":[67],"$1/\u03b5$":[68,146],"guarantees":[69],"scale":[70],"$H^5$":[72],"effective":[75],"horizon":[76,151],"$H$,":[77],"no":[79,123],"lower":[80],"bound":[81],"for":[82,95],"MDPs":[83,205],"depends":[84],"on":[85,203],"temperature":[87],"$\u03b2$.":[88],"In":[89],"work,":[91],"we":[92,190],"show":[93],"that":[94,109,122,127,183,196],"discounted":[97],"MDP":[98],"$K$":[100],"state-action":[101],"pairs,":[102],"plug-in":[104,165],"estimator":[105,176],"finds":[106],"$\u03b5$-optimal":[111],"every":[113,185],"initial":[114],"state":[115],"$\\widetilde{O}(KH^3/(\u03b5\\max\\{\u03b5,\u03b2\\})":[117],"+":[118],"KH)$":[119],"samples,":[120],"algorithm,":[124],"one":[126],"chooses":[128],"where":[129],"sample,":[131],"can":[132],"do":[133],"fewer":[135],"when":[136],"$\u03b5\\lesssim":[137],"\\min\\{H,":[138],"H^2/\u03b2\\}$.":[139],"For":[140],"$\u03b5<":[141],"\u03b2$":[142],"rate":[147],"same":[150],"dependence":[152],"unregularized":[154],"RL.":[155],"We":[156],"also":[157],"find":[158],"exact":[160],"asymptotic":[161],"constant":[162],"estimator,":[166],"Fisher-weighted":[168],"variance":[169],"estimated":[172],"soft":[173],"Q-function.":[174],"The":[175],"locally":[178],"asymptotically":[179],"minimax":[180],"among":[181,199],"methods":[182],"sample":[184],"pair":[186],"equally":[187],"often,":[188],"give":[191],"allocation":[193],"adaptive":[200],"designs.":[201],"Simulations":[202],"small":[204],"match":[206],"rates":[208],"constants.":[211]},"counts_by_year":[],"updated_date":"2026-11-12T12:28:57.059296","created_date":"2026-11-12T00:00:00"}