{"id":"https://openalex.org/W2098936573","doi":null,"title":"Signal-to-Noise Ratio Analysis of Policy Gradient Algorithms","display_name":"Signal-to-Noise Ratio Analysis of Policy Gradient Algorithms","publication_year":2008,"publication_date":"2008-12-08","ids":{"openalex":"https://openalex.org/W2098936573","mag":"2098936573"},"language":"en","primary_location":{"id":"pmh:oai:CiteSeerX.psu:10.1.1.208.5770","doi":null,"is_oa":false,"landing_page_url":"http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.208.5770","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"http://groups.csail.mit.edu/robotics-center/public_papers/Roberts08a.pdf","raw_type":"text"},"volume":null,"type":"article","is_traditional":true,"indexed_in":[],"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/A5000226143","display_name":"John W. Roberts","orcid":"https://orcid.org/0000-0002-0870-952X","observed_orcids":["https://orcid.org/0000-0002-0870-952X"]},"institutions":[{"id":"https://openalex.org/I4405258931","display_name":"MIT Computer Science and Artificial Intelligence Laboratory","ror":"https://ror.org/01e85r662","country_code":"US","type":"facility","lineage":["https://openalex.org/I4405258931","https://openalex.org/I63966007"]},{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"John W. Roberts","raw_affiliation_strings":["Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA#TAB#"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA#TAB#","institution_ids":["https://openalex.org/I4405258931","https://openalex.org/I63966007"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5074291890","display_name":"Russ Tedrake","orcid":"https://orcid.org/0000-0002-8712-7092","observed_orcids":["https://orcid.org/0000-0002-8712-7092"]},"institutions":[{"id":"https://openalex.org/I4405258931","display_name":"MIT Computer Science and Artificial Intelligence Laboratory","ror":"https://ror.org/01e85r662","country_code":"US","type":"facility","lineage":["https://openalex.org/I4405258931","https://openalex.org/I63966007"]},{"id":"https://openalex.org/I63966007","display_name":"Massachusetts Institute of Technology","ror":"https://ror.org/042nb2s44","country_code":"US","type":"education","lineage":["https://openalex.org/I63966007"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Russ Tedrake","raw_affiliation_strings":["Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA#TAB#"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, MA#TAB#","institution_ids":["https://openalex.org/I4405258931","https://openalex.org/I63966007"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":1.3961,"has_fulltext":false,"cited_by_count":14,"citation_normalized_percentile":{"value":0.84718124,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":88,"max":97},"biblio":{"volume":"21","issue":null,"first_page":"1361","last_page":"1368"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10462","display_name":"Reinforcement Learning in Robotics","score":1.0,"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"}},"topics":[{"id":"https://openalex.org/T10462","display_name":"Reinforcement Learning in Robotics","score":1.0,"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"}},{"id":"https://openalex.org/T12794","display_name":"Adaptive Dynamic Programming Control","score":0.0,"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-gradient","display_name":"policy gradient","score":0.8330000042915344},{"id":"https://openalex.org/keywords/signal-to-noise-ratio","display_name":"signal-to-noise ratio","score":0.9959999918937683},{"id":"https://openalex.org/keywords/gradient-variance-reduction","display_name":"gradient variance reduction","score":0.7649999856948853},{"id":"https://openalex.org/keywords/weight-perturbation","display_name":"weight perturbation","score":0.8379999995231628},{"id":"https://openalex.org/keywords/natural-gradient","display_name":"natural gradient","score":0.878000020980835},{"id":"https://openalex.org/keywords/shell-distribution","display_name":"shell distribution","score":0.6589999794960022},{"id":"https://openalex.org/keywords/model-free-reinforcement-learning","display_name":"model-free reinforcement learning","score":0.8059999942779541},{"id":"https://openalex.org/keywords/learning-performance-prediction","display_name":"learning outcome prediction","score":0.8349999785423279}],"concepts":[{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.6679354310035706},{"id":"https://openalex.org/C163716315","wikidata":"https://www.wikidata.org/wiki/Q901177","display_name":"Gaussian","level":2,"score":0.5247179269790649},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5168790817260742},{"id":"https://openalex.org/C62644790","wikidata":"https://www.wikidata.org/wiki/Q3454689","display_name":"Variance reduction","level":3,"score":0.5091196894645691},{"id":"https://openalex.org/C2777303404","wikidata":"https://www.wikidata.org/wiki/Q759757","display_name":"Convergence (economics)","level":2,"score":0.5029799342155457},{"id":"https://openalex.org/C52740198","wikidata":"https://www.wikidata.org/wiki/Q1539564","display_name":"Importance sampling","level":3,"score":0.45482826232910156},{"id":"https://openalex.org/C97541855","wikidata":"https://www.wikidata.org/wiki/Q830687","display_name":"Reinforcement learning","level":2,"score":0.4496048092842102},{"id":"https://openalex.org/C13944312","wikidata":"https://www.wikidata.org/wiki/Q7512748","display_name":"Signal-to-noise ratio (imaging)","level":2,"score":0.4494287371635437},{"id":"https://openalex.org/C115680565","wikidata":"https://www.wikidata.org/wiki/Q5977448","display_name":"Gradient method","level":2,"score":0.4418971538543701},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.39608117938041687},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.37035995721817017},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.29941657185554504},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.28251445293426514},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.08731278777122498},{"id":"https://openalex.org/C19499675","wikidata":"https://www.wikidata.org/wiki/Q232207","display_name":"Monte Carlo method","level":2,"score":0.08684167265892029},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C50522688","wikidata":"https://www.wikidata.org/wiki/Q189833","display_name":"Economic growth","level":1,"score":0.0},{"id":"https://openalex.org/C162324750","wikidata":"https://www.wikidata.org/wiki/Q8134","display_name":"Economics","level":0,"score":0.0}],"mesh":[],"locations_count":4,"locations":[{"id":"pmh:oai:CiteSeerX.psu:10.1.1.208.5770","doi":null,"is_oa":false,"landing_page_url":"http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.208.5770","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"http://groups.csail.mit.edu/robotics-center/public_papers/Roberts08a.pdf","raw_type":"text"},{"id":"pmh:oai:CiteSeerX.psu:10.1.1.566.9181","doi":null,"is_oa":false,"landing_page_url":"http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.566.9181","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"http://books.nips.cc/papers/files/nips21/NIPS2008_0715.pdf","raw_type":"text"},{"id":"pmh:oai:CiteSeerX.psu:10.1.1.855.2009","doi":null,"is_oa":false,"landing_page_url":"http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.855.2009","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"http://papers.nips.cc/paper/3511-signal-to-noise-ratio-analysis-of-policy-gradient-algorithms.pdf","raw_type":"text"},{"id":"mag:2098936573","doi":null,"is_oa":false,"landing_page_url":"http://groups.csail.mit.edu/robotics-center/public_papers/Roberts08a.pdf","pdf_url":null,"source":{"id":"https://openalex.org/S4363606243","display_name":"neural information processing systems","issn_l":"1049-5258","issn":["1049-5258"],"is_oa":false,"is_in_doaj":false,"is_core":true,"listed_in":["cwts-core","norway-2"],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":"Neural Information Processing Systems","raw_type":null}],"best_oa_location":null,"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":15,"referenced_works":["https://openalex.org/W55510407","https://openalex.org/W1590243462","https://openalex.org/W1594783240","https://openalex.org/W1970789124","https://openalex.org/W2041301661","https://openalex.org/W2056318256","https://openalex.org/W2108682071","https://openalex.org/W2109008048","https://openalex.org/W2119717200","https://openalex.org/W2126330148","https://openalex.org/W2130801532","https://openalex.org/W2139053308","https://openalex.org/W2144446635","https://openalex.org/W2146737184","https://openalex.org/W2156084351"],"related_works":["https://openalex.org/W7127319615","https://openalex.org/W2285432433","https://openalex.org/W2116383462","https://openalex.org/W2974544365","https://openalex.org/W7098082322","https://openalex.org/W2029278774","https://openalex.org/W7101199588","https://openalex.org/W3026994642","https://openalex.org/W2152238966","https://openalex.org/W2382229373"],"abstract_inverted_index":{"Policy":[0],"gradient":[1,45,94,130,181],"(PG)":[2],"reinforcement":[3],"learning":[4,13,176],"algorithms":[5,95],"have":[6],"strong":[7],"(local)":[8],"con-vergence":[9],"guarantees,":[10],"but":[11,117],"their":[12],"performance":[14,177],"is":[15,63,80,153],"typically":[16],"limited":[17],"by":[18,37],"a":[19,39,64,112,154,159],"large":[20],"variance":[21,34],"in":[22,73,96,165,175,178],"the":[23,26,33,53,77,82,100,115,119,128,132,144,149],"estimate":[24],"of":[25,67,131],"gradient.":[27],"In":[28],"this":[29,49,121],"paper,":[30],"we":[31,103,136],"formulate":[32],"reduction":[35],"problem":[36],"describing":[38],"signal-to-noise":[40],"ratio":[41],"(SNR)":[42],"for":[43,52],"policy":[44,93,180],"algorithms,":[46],"and":[47,71,146,162],"evaluate":[48],"SNR":[50,62],"carefully":[51],"popular":[54],"Weight":[55],"Perturbation":[56],"(WP)":[57],"algorithm.":[58],"We":[59,85,167],"confirm":[60],"that":[61,72,138,148,169],"good":[65],"predictor":[66],"long-term":[68],"learn-ing":[69],"performance,":[70],"our":[74],"episodic":[75],"formulation,":[76],"cost-to-go":[78],"function":[79],"indeed":[81],"optimal":[83,150],"baseline.":[84],"then":[86],"propose":[87],"two":[88],"modifications":[89,171],"to":[90,98],"traditional":[91],"model-free":[92],"order":[97],"optimize":[99],"SNR.":[101],"First,":[102],"examine":[104],"WP":[105],"using":[106],"anisotropic":[107],"sampling":[108],"distributions,":[109],"which":[110],"introduces":[111],"bias":[113,122],"into":[114],"update":[116],"increases":[118],"SNR;":[120],"can":[123,141],"be":[124],"interpreted":[125],"as":[126],"following":[127],"natural":[129],"cost":[133],"function.":[134],"Second,":[135],"show":[137],"non-Gaussian":[139],"distribu-tions":[140],"also":[142],"increase":[143],"SNR,":[145],"argue":[147],"isotropic":[151],"distribution":[152,157,164],"\u2018shell":[155],"\u2019":[156],"with":[158],"constant":[160],"magnitude":[161],"uniform":[163],"direc-tion.":[166],"demonstrate":[168],"both":[170],"produce":[172],"substantial":[173],"improvements":[174],"challenging":[179],"experiments.":[182],"1":[183]},"counts_by_year":[{"year":2023,"cited_by_count":1},{"year":2020,"cited_by_count":1},{"year":2019,"cited_by_count":3},{"year":2018,"cited_by_count":1},{"year":2017,"cited_by_count":1},{"year":2014,"cited_by_count":1},{"year":2013,"cited_by_count":2},{"year":2012,"cited_by_count":1}],"updated_date":"2026-11-04T10:54:05.112534","created_date":"2016-06-24T00:00:00"}