{"id":"https://openalex.org/W4408355490","doi":"https://doi.org/10.1109/icassp49660.2025.10889234","title":"Variance-reduced Clipping for Non-convex Optimization","display_name":"Variance-reduced Clipping for Non-convex Optimization","publication_year":2025,"publication_date":"2025-03-12","ids":{"openalex":"https://openalex.org/W4408355490","doi":"https://doi.org/10.1109/icassp49660.2025.10889234"},"language":"en","primary_location":{"id":"doi:10.1109/icassp49660.2025.10889234","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49660.2025.10889234","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2025 - 2025 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/A5059780399","display_name":"Amirhossein Reisizadeh","orcid":"https://orcid.org/0000-0002-1730-8402"},"institutions":[{"id":"https://openalex.org/I4210110987","display_name":"IIT@MIT","ror":"https://ror.org/01wp8zh54","country_code":"US","type":"facility","lineage":["https://openalex.org/I30771326","https://openalex.org/I4210110987"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Amirhossein Reisizadeh","raw_affiliation_strings":["MIT,LIDS,Cambridge,MA,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT,LIDS,Cambridge,MA,USA","institution_ids":["https://openalex.org/I4210110987"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5024083506","display_name":"Haochuan Li","orcid":"https://orcid.org/0000-0002-4857-6709"},"institutions":[{"id":"https://openalex.org/I4210110987","display_name":"IIT@MIT","ror":"https://ror.org/01wp8zh54","country_code":"US","type":"facility","lineage":["https://openalex.org/I30771326","https://openalex.org/I4210110987"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Haochuan Li","raw_affiliation_strings":["MIT,LIDS,Cambridge,MA,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT,LIDS,Cambridge,MA,USA","institution_ids":["https://openalex.org/I4210110987"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5011183160","display_name":"Subhro Das","orcid":"https://orcid.org/0000-0002-7610-2738"},"institutions":[{"id":"https://openalex.org/I4210087032","display_name":"Cambridge Scientific (United States)","ror":"https://ror.org/001s4dh65","country_code":"US","type":"company","lineage":["https://openalex.org/I4210087032"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Subhro Das","raw_affiliation_strings":["IBM Research,Cambridge,MA,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"IBM Research,Cambridge,MA,USA","institution_ids":["https://openalex.org/I4210087032"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5078288116","display_name":"Ali Jadbabaie","orcid":"https://orcid.org/0000-0003-1122-3069"},"institutions":[{"id":"https://openalex.org/I4210110987","display_name":"IIT@MIT","ror":"https://ror.org/01wp8zh54","country_code":"US","type":"facility","lineage":["https://openalex.org/I30771326","https://openalex.org/I4210110987"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Ali Jadbabaie","raw_affiliation_strings":["MIT,LIDS,Cambridge,MA,USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"MIT,LIDS,Cambridge,MA,USA","institution_ids":["https://openalex.org/I4210110987"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"5"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11749","display_name":"Iterative Learning Control Systems","score":0.8928999900817871,"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"}},"topics":[{"id":"https://openalex.org/T11749","display_name":"Iterative Learning Control Systems","score":0.8928999900817871,"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/T10791","display_name":"Advanced Control Systems Optimization","score":0.7803999781608582,"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/T10551","display_name":"Scheduling and Optimization Algorithms","score":0.7276999950408936,"subfield":{"id":"https://openalex.org/subfields/2209","display_name":"Industrial and Manufacturing Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/clipping","display_name":"Clipping (morphology)","score":0.6496198773384094},{"id":"https://openalex.org/keywords/regular-polygon","display_name":"Regular polygon","score":0.4891989231109619},{"id":"https://openalex.org/keywords/variance","display_name":"Variance (accounting)","score":0.4890042543411255},{"id":"https://openalex.org/keywords/convex-optimization","display_name":"Convex optimization","score":0.47982001304626465},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.4689673185348511},{"id":"https://openalex.org/keywords/mathematical-optimization","display_name":"Mathematical optimization","score":0.37717923521995544},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.32915687561035156},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.3200618624687195},{"id":"https://openalex.org/keywords/geometry","display_name":"Geometry","score":0.07108679413795471}],"concepts":[{"id":"https://openalex.org/C2776848632","wikidata":"https://www.wikidata.org/wiki/Q853463","display_name":"Clipping (morphology)","level":2,"score":0.6496198773384094},{"id":"https://openalex.org/C112680207","wikidata":"https://www.wikidata.org/wiki/Q714886","display_name":"Regular polygon","level":2,"score":0.4891989231109619},{"id":"https://openalex.org/C196083921","wikidata":"https://www.wikidata.org/wiki/Q7915758","display_name":"Variance (accounting)","level":2,"score":0.4890042543411255},{"id":"https://openalex.org/C157972887","wikidata":"https://www.wikidata.org/wiki/Q463359","display_name":"Convex optimization","level":3,"score":0.47982001304626465},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.4689673185348511},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.37717923521995544},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.32915687561035156},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3200618624687195},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.07108679413795471},{"id":"https://openalex.org/C144133560","wikidata":"https://www.wikidata.org/wiki/Q4830453","display_name":"Business","level":0,"score":0.0},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.0},{"id":"https://openalex.org/C121955636","wikidata":"https://www.wikidata.org/wiki/Q4116214","display_name":"Accounting","level":1,"score":0.0},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icassp49660.2025.10889234","is_oa":false,"landing_page_url":"https://doi.org/10.1109/icassp49660.2025.10889234","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"ICASSP 2025 - 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":20,"referenced_works":["https://openalex.org/W2194775991","https://openalex.org/W2962739339","https://openalex.org/W2993258424","https://openalex.org/W4408355490","https://openalex.org/W6675761068","https://openalex.org/W6680196509","https://openalex.org/W6698038680","https://openalex.org/W6698102009","https://openalex.org/W6737778391","https://openalex.org/W6740708716","https://openalex.org/W6742632731","https://openalex.org/W6753122105","https://openalex.org/W6757358033","https://openalex.org/W6763249885","https://openalex.org/W6764647576","https://openalex.org/W6772129282","https://openalex.org/W6783674534","https://openalex.org/W6791321639","https://openalex.org/W6794728472","https://openalex.org/W6842319320"],"related_works":["https://openalex.org/W2035130982","https://openalex.org/W3198205105","https://openalex.org/W2956000802","https://openalex.org/W3198227724","https://openalex.org/W110878719","https://openalex.org/W2098726873","https://openalex.org/W3183431084","https://openalex.org/W1523583400","https://openalex.org/W2122112927","https://openalex.org/W2811267387"],"abstract_inverted_index":{"Gradient":[0],"clipping":[1,116],"is":[2,55,72,92,155,166,172,204],"a":[3,26,78,93,138,148],"standard":[4],"training":[5,35],"technique":[6],"used":[7],"in":[8,30,56,63,101],"deep":[9],"learning":[10,151],"applications":[11],"such":[12,99],"as":[13,206],"large-scale":[14],"language":[15],"modeling":[16],"to":[17,59,74,128,157,191],"mitigate":[18],"exploding":[19],"gradients.":[20],"Recent":[21],"experimental":[22],"studies":[23],"have":[24],"demonstrated":[25],"fairly":[27],"special":[28],"behavior":[29,100],"the":[31,34,47,51,60,70,170,173,180],"smoothness":[32,48,71,91,112],"of":[33,175],"objective":[36,171],"along":[37],"its":[38],"trajectory":[39],"when":[40,169],"trained":[41],"with":[42,50,115],"gradient":[43,52,126,189],"clipping.":[44],"That":[45],"is,":[46],"grows":[49],"norm.":[53],"This":[54],"clear":[57],"contrast":[58],"wellestablished":[61],"assumption":[62],"folklore":[64],"non-convex":[65,102],"optimization,":[66],"a.k.a.":[67],"L\u2013smoothness,":[68],"where":[69],"assumed":[73],"be":[75],"bounded":[76],"by":[77],"constant":[79],"L":[80],"globally.":[81],"The":[82],"recently":[83],"introduced":[84],"(L<inf":[85],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[86,89],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">0</inf>,":[87],"L<inf":[88],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">1</inf>)\u2013":[90],"more":[94],"relaxed":[95,111],"notion":[96],"that":[97,108,146],"captures":[98],"optimization.":[103],"It":[104],"has":[105],"been":[106],"shown":[107],"under":[109],"this":[110,134,153],"assumption,":[113],"SGD":[114],"requires":[117],"$\\mathcal{O}\\left(":[118,158,182,192],"{{":[119,159],"\\in":[120,160,184,196],"^{":[121,161,185,197],"-":[122,162,186,198],"4}}}":[123],"\\right)$":[124,164,188],"stochastic":[125],"computations":[127],"find":[129],"an":[130],"\u03f5\u2013stationary":[131],"solution.":[132],"In":[133],"paper,":[135],"we":[136,178],"employ":[137],"variance":[139],"reduction":[140],"technique,":[141],"namely":[142],"Spider,":[143],"and":[144],"demonstrate":[145],"for":[147],"carefully":[149],"designed":[150],"rate,":[152],"complexity":[154,190],"improved":[156],"3}}}":[163],"which":[165,203],"order-optimal.":[167],"Moreover,":[168],"average":[174],"n":[176,194],"components,":[177],"improve":[179],"existing":[181],"{n{":[183],"2}}}":[187],"{\\sqrt":[193],"{":[195],"2}}":[199],"+":[200],"n}":[201],"\\right)$,":[202],"order-optimal":[205],"well.":[207]},"counts_by_year":[{"year":2025,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}