{"id":"https://openalex.org/W3160355689","doi":"https://doi.org/10.1109/icpr48806.2021.9413027","title":"The Effect of Multi-step Methods on Overestimation in Deep Reinforcement Learning","display_name":"The Effect of Multi-step Methods on Overestimation in Deep Reinforcement Learning","publication_year":2021,"publication_date":"2021-01-10","ids":{"openalex":"https://openalex.org/W3160355689","doi":"https://doi.org/10.1109/icpr48806.2021.9413027","mag":"3160355689"},"language":"en","primary_location":{"id":"doi:10.1109/icpr48806.2021.9413027","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icpr48806.2021.9413027","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2020 25th International Conference on Pattern Recognition (ICPR)","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":false,"oa_status":"closed","oa_url":null,"any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5048233348","display_name":"Lingheng Meng","orcid":"https://orcid.org/0000-0001-9419-5774"},"institutions":[{"id":"https://openalex.org/I151746483","display_name":"University of Waterloo","ror":"https://ror.org/01aff2v68","country_code":"CA","type":"education","lineage":["https://openalex.org/I151746483"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Lingheng Meng","raw_affiliation_strings":["Electrical and Computer Engineering, University of Waterloo, Waterloo, Canada"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Electrical and Computer Engineering, University of Waterloo, Waterloo, Canada","institution_ids":["https://openalex.org/I151746483"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5036605802","display_name":"Rob Gorbet","orcid":"https://orcid.org/0000-0003-2799-3922"},"institutions":[{"id":"https://openalex.org/I151746483","display_name":"University of Waterloo","ror":"https://ror.org/01aff2v68","country_code":"CA","type":"education","lineage":["https://openalex.org/I151746483"]}],"countries":["CA"],"is_corresponding":false,"raw_author_name":"Rob Gorbet","raw_affiliation_strings":["Knowledge Integration, University of Waterloo, Waterloo, Canada"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Knowledge Integration, University of Waterloo, Waterloo, Canada","institution_ids":["https://openalex.org/I151746483"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5072377971","display_name":"Dana Kuli\u0107","orcid":"https://orcid.org/0000-0002-4169-2141"},"institutions":[{"id":"https://openalex.org/I56590836","display_name":"Monash University","ror":"https://ror.org/02bfwt286","country_code":"AU","type":"education","lineage":["https://openalex.org/I56590836"]}],"countries":["AU"],"is_corresponding":false,"raw_author_name":"Dana Kulic","raw_affiliation_strings":["Monash University, Melbourne, Australia"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Monash University, Melbourne, Australia","institution_ids":["https://openalex.org/I56590836"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":2.4567,"has_fulltext":false,"cited_by_count":27,"citation_normalized_percentile":{"value":0.90587776,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":89,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"347","last_page":"353"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10462","display_name":"Reinforcement Learning in Robotics","score":0.9988999962806702,"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":0.9988999962806702,"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/reinforcement-learning","display_name":"Reinforcement learning","score":0.9142586588859558},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7999697923660278},{"id":"https://openalex.org/keywords/backup","display_name":"Backup","score":0.6066620945930481},{"id":"https://openalex.org/keywords/set","display_name":"Set (abstract data type)","score":0.5405243039131165},{"id":"https://openalex.org/keywords/function","display_name":"Function (biology)","score":0.5366707444190979},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.523280918598175},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.5026185512542725},{"id":"https://openalex.org/keywords/bellman-equation","display_name":"Bellman equation","score":0.4987297058105469},{"id":"https://openalex.org/keywords/two-step","display_name":"Two step","score":0.4887402057647705},{"id":"https://openalex.org/keywords/function-approximation","display_name":"Function approximation","score":0.4425974488258362},{"id":"https://openalex.org/keywords/value","display_name":"Value (mathematics)","score":0.42569243907928467},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.40574532747268677},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.3996732234954834},{"id":"https://openalex.org/keywords/mathematical-optimization","display_name":"Mathematical optimization","score":0.32250723242759705},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.13102957606315613},{"id":"https://openalex.org/keywords/applied-mathematics","display_name":"Applied mathematics","score":0.07263895869255066}],"concepts":[{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.9142586588859558},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7999697923660278},{"id":"https://openalex.org/C2780945871","wikidata":"https://www.wikidata.org/wiki/Q194274","display_name":"Backup","level":2,"score":0.6066620945930481},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.5405243039131165},{"id":"https://openalex.org/C14036430","wikidata":"https://www.wikidata.org/wiki/Q3736076","display_name":"Function (biology)","level":2,"score":0.5366707444190979},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.523280918598175},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.5026185512542725},{"id":"https://openalex.org/C14646407","wikidata":"https://www.wikidata.org/wiki/Q1430750","display_name":"Bellman equation","level":2,"score":0.4987297058105469},{"id":"https://openalex.org/C3019136120","wikidata":"https://www.wikidata.org/wiki/Q2462428","display_name":"Two step","level":2,"score":0.4887402057647705},{"id":"https://openalex.org/C91873725","wikidata":"https://www.wikidata.org/wiki/Q3445816","display_name":"Function approximation","level":3,"score":0.4425974488258362},{"id":"https://openalex.org/C2776291640","wikidata":"https://www.wikidata.org/wiki/Q2912517","display_name":"Value (mathematics)","level":2,"score":0.42569243907928467},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.40574532747268677},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.3996732234954834},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.32250723242759705},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.13102957606315613},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.07263895869255066},{"id":"https://openalex.org/C78458016","wikidata":"https://www.wikidata.org/wiki/Q840400","display_name":"Evolutionary biology","level":1,"score":0.0},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.0},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.0},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icpr48806.2021.9413027","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icpr48806.2021.9413027","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2020 25th International Conference on Pattern Recognition (ICPR)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320334617","display_name":"Social Sciences and Humanities Research Council of Canada","ror":"https://ror.org/04j5jqy92"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":46,"referenced_works":["https://openalex.org/W51508254","https://openalex.org/W2115211925","https://openalex.org/W2121863487","https://openalex.org/W2145339207","https://openalex.org/W2155968351","https://openalex.org/W2158782408","https://openalex.org/W2165150801","https://openalex.org/W2173248099","https://openalex.org/W2201581102","https://openalex.org/W2518491366","https://openalex.org/W2560647685","https://openalex.org/W2593237273","https://openalex.org/W2596758708","https://openalex.org/W2623491082","https://openalex.org/W2746553466","https://openalex.org/W2754517384","https://openalex.org/W2761873684","https://openalex.org/W2781726626","https://openalex.org/W2786928559","https://openalex.org/W2787938642","https://openalex.org/W2789824229","https://openalex.org/W2810754397","https://openalex.org/W2912496897","https://openalex.org/W2937599623","https://openalex.org/W2953364219","https://openalex.org/W2963120839","https://openalex.org/W2963477884","https://openalex.org/W2963828709","https://openalex.org/W2963864421","https://openalex.org/W2964001908","https://openalex.org/W2964043796","https://openalex.org/W2964248288","https://openalex.org/W2964291307","https://openalex.org/W3037207827","https://openalex.org/W4214717370","https://openalex.org/W4300799055","https://openalex.org/W4302570325","https://openalex.org/W6602057636","https://openalex.org/W6677067356","https://openalex.org/W6692846177","https://openalex.org/W6735506055","https://openalex.org/W6740801417","https://openalex.org/W6744123322","https://openalex.org/W6744838376","https://openalex.org/W6750253780","https://openalex.org/W6753264383"],"related_works":["https://openalex.org/W2386410636","https://openalex.org/W3038962357","https://openalex.org/W2025663273","https://openalex.org/W3099153698","https://openalex.org/W1501190258","https://openalex.org/W3021988786","https://openalex.org/W4256087190","https://openalex.org/W3099311996","https://openalex.org/W4240668504","https://openalex.org/W2788366696"],"abstract_inverted_index":{"Multi-step":[0,142],"(also":[1],"called":[2,140],"n-step)":[3],"methods":[4,51,97],"in":[5,32,42,58,103,186,207,254],"Reinforcement":[6,44],"Learning":[7,45],"(RL)":[8],"have":[9],"been":[10],"shown":[11],"to":[12,21,83,203,209,251],"be":[13],"more":[14],"efficient":[15],"than":[16,178],"the":[17,25,38,61,84,93,100,159,175,196,215,229],"1-step":[18,181],"method":[19],"due":[20],"faster":[22],"propagation":[23],"of":[24,37,76,86,95,118,200,233,239],"reward":[26],"signal,":[27],"both":[28,166],"theoretically":[29],"and":[30,55,63,137,168,190,198,213,219,235,263],"empirically,":[31],"tasks":[33],"exploiting":[34],"tabular":[35],"representation":[36],"value-function.":[39],"Recently,":[40],"research":[41],"Deep":[43,119,242],"(DRL)":[46],"also":[47,194],"shows":[48],"that":[49,165,221],"multi-step":[50,96,106,150,205,224],"improve":[52],"learning":[53,191,264],"speed":[54],"final":[56,188,261],"performance":[57,189,262],"applications":[59],"where":[60,105,130,145],"value-function":[62],"policy":[64],"are":[65,108,134,170],"represented":[66],"with":[67,180,237],"deep":[68],"neural":[69],"networks.":[70],"However,":[71],"there":[72],"is":[73,80,152],"a":[74,111,138,247],"lack":[75],"understanding":[77],"about":[78],"what":[79],"actually":[81],"contributing":[82],"boost":[85],"performance.":[87],"In":[88],"this":[89],"work,":[90],"we":[91,124,163,227],"analyze":[92],"effect":[94],"on":[98,116],"alleviating":[99],"overestimation":[101,176,218,253],"problem":[102,177],"DRL,":[104],"experiences":[107],"sampled":[109],"from":[110],"replay":[112],"buffer.":[113],"Specifically":[114],"building":[115],"top":[117],"Deterministic":[120,243],"Policy":[121,244],"Gradient":[122,245],"(DDPG),":[123],"propose":[125],"Multi-":[126],"step":[127,132],"DDPG":[128,143,179],"(MDDPG),":[129],"different":[131,149,201],"sizes":[133],"manually":[135],"set,":[136],"variant":[139],"Mixed":[141],"(MMDDPG)":[144],"an":[146,155],"average":[147],"over":[148],"backups":[151],"used":[153],"as":[154],"update":[156],"target":[157],"for":[158],"Q-value":[160],"function.":[161],"Empirically,":[162],"show":[164,259],"MDDPG":[167,234],"MMDDPG":[169,236],"significantly":[171],"less":[172],"affected":[173],"by":[174],"backup,":[182],"which":[183],"consequently":[184],"results":[185],"better":[187],"speed.":[192,265],"We":[193],"discuss":[195],"advantages":[197],"disadvantages":[199],"ways":[202],"do":[204],"expansion":[206],"order":[208],"reduce":[210],"approximation":[211],"error,":[212],"expose":[214],"tradeoff":[216],"between":[217],"underestimation":[220],"underlies":[222],"offline":[223],"methods.":[225],"Finally,":[226],"compare":[228],"computational":[230],"resource":[231],"needs":[232],"those":[238],"Twin":[240],"Delayed":[241],"(TD3),":[246],"state-of-the-art":[248],"algorithm":[249],"proposed":[250],"address":[252],"actor-critic":[255],"methods,":[256],"since":[257],"they":[258],"comparable":[260]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2025,"cited_by_count":7},{"year":2024,"cited_by_count":3},{"year":2023,"cited_by_count":8},{"year":2022,"cited_by_count":5},{"year":2021,"cited_by_count":2},{"year":2020,"cited_by_count":1}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}