{"id":"https://openalex.org/W4372260246","doi":"https://doi.org/10.1109/icassp49357.2023.10096624","title":"Improving the Stochastic Gradient Descent\u2019s Test Accuracy by Manipulating the \u2113<sub>\u221e</sub> Norm of its Gradient Approximation","display_name":"Improving the Stochastic Gradient Descent\u2019s Test Accuracy by Manipulating the \u2113<sub>\u221e</sub> Norm of its Gradient Approximation","publication_year":2023,"publication_date":"2023-05-05","ids":{"openalex":"https://openalex.org/W4372260246","doi":"https://doi.org/10.1109/icassp49357.2023.10096624"},"language":"en","primary_location":{"id":"doi:10.1109/icassp49357.2023.10096624","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/icassp49357.2023.10096624","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"},"type":"conference-paper","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/A5077092609","display_name":"Paul Rodr\u00edguez","orcid":"https://orcid.org/0000-0002-8501-0907"},"institutions":[{"id":"https://openalex.org/I65285256","display_name":"Pontificia Universidad Cat\u00f3lica del Per\u00fa","ror":"https://ror.org/00013q465","country_code":"PE","type":"education","lineage":["https://openalex.org/I65285256"]}],"countries":["PE"],"is_corresponding":true,"raw_author_name":"Paul Rodriguez","raw_affiliation_strings":["Pontificia Universidad Cat&#x00F3;lica del Per&#x00FA;,Electrical Department,Lima,Peru"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Pontificia Universidad Cat&#x00F3;lica del Per&#x00FA;,Electrical Department,Lima,Peru","institution_ids":["https://openalex.org/I65285256"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":["https://openalex.org/A5077092609"],"corresponding_institution_ids":["https://openalex.org/I65285256"],"apc_list":null,"apc_paid":null,"fwci":0.7477,"has_fulltext":false,"cited_by_count":2,"citation_normalized_percentile":{"value":0.76023098,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":94,"max":96},"biblio":{"volume":"27","issue":null,"first_page":"1","last_page":"5"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10500","display_name":"Sparse and Compressive Sensing Techniques","score":0.9979000091552734,"subfield":{"id":"https://openalex.org/subfields/2206","display_name":"Computational Mechanics"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T10500","display_name":"Sparse and Compressive Sensing Techniques","score":0.9979000091552734,"subfield":{"id":"https://openalex.org/subfields/2206","display_name":"Computational Mechanics"},"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.9979000091552734,"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/T10963","display_name":"Advanced Optimization Algorithms Research","score":0.9871000051498413,"subfield":{"id":"https://openalex.org/subfields/2612","display_name":"Numerical Analysis"},"field":{"id":"https://openalex.org/fields/26","display_name":"Mathematics"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/stochastic-gradient-descent","display_name":"Stochastic gradient descent","score":0.7343419194221497},{"id":"https://openalex.org/keywords/gradient-descent","display_name":"Gradient descent","score":0.6556086540222168},{"id":"https://openalex.org/keywords/norm","display_name":"Norm (philosophy)","score":0.5795860886573792},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.44759610295295715},{"id":"https://openalex.org/keywords/applied-mathematics","display_name":"Applied mathematics","score":0.4417431056499481},{"id":"https://openalex.org/keywords/gradient-method","display_name":"Gradient method","score":0.4160751700401306},{"id":"https://openalex.org/keywords/mathematical-optimization","display_name":"Mathematical optimization","score":0.3539919853210449},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.35078489780426025},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.32150495052337646},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.17213883996009827},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.07907754182815552}],"concepts":[{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.7343419194221497},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.6556086540222168},{"id":"https://openalex.org/C191795146","wikidata":"https://www.wikidata.org/wiki/Q3878446","display_name":"Norm (philosophy)","level":2,"score":0.5795860886573792},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.44759610295295715},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.4417431056499481},{"id":"https://openalex.org/C115680565","wikidata":"https://www.wikidata.org/wiki/Q5977448","display_name":"Gradient method","level":2,"score":0.4160751700401306},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.3539919853210449},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.35078489780426025},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.32150495052337646},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.17213883996009827},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.07907754182815552},{"id":"https://openalex.org/C17744445","wikidata":"https://www.wikidata.org/wiki/Q36442","display_name":"Political science","level":0,"score":0.0},{"id":"https://openalex.org/C199539241","wikidata":"https://www.wikidata.org/wiki/Q7748","display_name":"Law","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp49357.2023.10096624","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/icassp49357.2023.10096624","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[{"id":"https://openalex.org/F4320338281","display_name":"Army Research Office","ror":"https://ror.org/05epdh915"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":38,"referenced_works":["https://openalex.org/W6908809","https://openalex.org/W1522301498","https://openalex.org/W1686810756","https://openalex.org/W1955857676","https://openalex.org/W1988720110","https://openalex.org/W1989098643","https://openalex.org/W2006262045","https://openalex.org/W2031906930","https://openalex.org/W2032483793","https://openalex.org/W2058532290","https://openalex.org/W2100556411","https://openalex.org/W2119058682","https://openalex.org/W2142224912","https://openalex.org/W2146502635","https://openalex.org/W2148825261","https://openalex.org/W2163605009","https://openalex.org/W2523246573","https://openalex.org/W2896101215","https://openalex.org/W2936995161","https://openalex.org/W2949935872","https://openalex.org/W2964094444","https://openalex.org/W2988640914","https://openalex.org/W3100197791","https://openalex.org/W3133262835","https://openalex.org/W3175674388","https://openalex.org/W4205314337","https://openalex.org/W4241652050","https://openalex.org/W4292363360","https://openalex.org/W6631190155","https://openalex.org/W6637373629","https://openalex.org/W6681435938","https://openalex.org/W6681723013","https://openalex.org/W6684191040","https://openalex.org/W6727249380","https://openalex.org/W6755977018","https://openalex.org/W6757751764","https://openalex.org/W6761030284","https://openalex.org/W6784591795"],"related_works":["https://openalex.org/W4206903459","https://openalex.org/W2754816816","https://openalex.org/W4366280654","https://openalex.org/W3160167280","https://openalex.org/W4231621013","https://openalex.org/W4362706668","https://openalex.org/W3008318776","https://openalex.org/W2041416246","https://openalex.org/W3020853991","https://openalex.org/W3035836947"],"abstract_inverted_index":{"The":[0],"stochastic":[1],"gradient":[2,119,145],"descent":[3],"(SGD)":[4],"is":[5,23,70],"a":[6,18,49,71,214],"simple":[7,83],"yet":[8],"very":[9],"influential":[10],"algorithm":[11],"used":[12],"to":[13,89,103,134,160,166,177,203,219,236],"find":[14],"the":[15,44,90,118,142,151],"minimum":[16],"of":[17,67,79,148],"loss":[19],"(cost)":[20],"function":[21],"which":[22,47,76],"dependent":[24],"on":[25,132],"datasets":[26],"with":[27,35,113,190],"large":[28],"cardinality,":[29],"such":[30,85,106],"in":[31],"cases":[32],"typically":[33],"associated":[34],"deep":[36],"learning":[37],"(DL).":[38],"There":[39],"exists":[40],"several":[41,191],"variants/improvements":[42],"over":[43],"\"vanilla\"":[45],"SGD,":[46],"from":[48,63,82],"high-level":[50],"perspective,":[51],"may":[52],"be":[53,178],"understood":[54],"as":[55,165,199],"using":[56,150],"an":[57,64,97],"adaptive":[58],"element-wise":[59],"step-size":[60],"(SS).":[61],"Moreover,":[62],"algorithmic":[65],"point":[66],"view,":[68],"there":[69],"clear":[72],"\"incremental":[73],"improvement":[74],"path\"":[75],"relates":[77],"all":[78],"them,":[80],"i.e.":[81],"alternative":[84,104],"SG":[86],"Clipping":[87],"(SGC)":[88],"well-known":[91,238],"variance":[92],"correction":[93,111],"(Adagrad),":[94],"follow":[95],"by":[96,128],"(EMA)":[98],"exponential":[99],"moving":[100,175],"average":[101,176],"(RMSprop)":[102],"furtherance":[105],"Newton":[107],"(AdaDelta)":[108],"or":[109],"bias":[110],"along":[112],"different":[114],"EMA":[115,155],"options":[116],"for":[117,137],"itself":[120],"(Adam,":[121],"AdaMAx,":[122],"AdaBelief,":[123],"etc.).In":[124],"this":[125],"paper,":[126],"inspired":[127],"previous":[129],"non-stochastic":[130],"results":[131,183],"how":[133],"avoid":[135],"divergence":[136,208],"ill":[138],"chosen":[139],"SS":[140,206],"(for":[141],"accelerated":[143],"proximal":[144],"algorithm),":[146],"instead":[147],"directly":[149],"standard":[152],"SGD":[153],"gradient\u2019s":[154],"${{\\mathbf{\\bar":[156],"g}}_k}$,":[157],"we":[158],"propose":[159],"modify":[161],"its":[162,226],"entries":[163],"so":[164],"force":[167],"$\\left\\{":[168],"{{{\\left\\|":[169],"{{{{\\mathbf{\\bar":[170],"g}}}_k}}":[171],"\\right\\|}_\\infty":[172],"}}":[173],"\\right\\}{\\text{'s}}$":[174],"non-increasing.":[179],"Our":[180],"reproducible":[181],"computational":[182],"compare":[184],"our":[185],"proposed":[186],"algorithm,":[187],"called":[188],"SGD-\u2113\u221e,":[189],"optimizers":[192],"(such":[193],"Adam,":[194],"AdaMax,":[195],"SGC,":[196],"etc.);":[197],"while,":[198],"expected,":[200],"SGD-\u2113\u221eallows":[201],"us":[202],"use":[204],"larger":[205],"without":[207],"problems,":[209],"(i)":[210],"it":[211],"also":[212],"matches":[213],"well-tuned":[215],"Adam\u2019s":[216],"performance":[217],"(superior":[218],"\"default":[220],"parameters\"":[221],"Adam),":[222],"and":[223],"(ii)":[224],"heuristically,":[225],"convergence":[227],"properties":[228],"(rate,":[229],"oscillations,":[230],"etc.)":[231],"are":[232],"superior":[233],"when":[234],"compared":[235],"other":[237],"algorithms.":[239]},"counts_by_year":[{"year":2024,"cited_by_count":2}],"updated_date":"2026-07-29T14:22:42.915294","created_date":"2025-10-10T00:00:00"}