{"id":"https://openalex.org/W2287409372","doi":null,"title":"Bilevel programming algorithms for machine learning model selection","display_name":"Bilevel programming algorithms for machine learning model selection","publication_year":2010,"publication_date":"2010-01-01","ids":{"openalex":"https://openalex.org/W2287409372","mag":"2287409372"},"language":"en","primary_location":{"id":"mag:2287409372","doi":null,"is_oa":false,"landing_page_url":"http://dl.acm.org/citation.cfm?id=2125745","pdf_url":null,"source":null,"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":null},"volume":null,"type":"article","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/A5048876983","display_name":"Kristin P. Bennett","orcid":"https://orcid.org/0000-0002-8782-105X","observed_orcids":["https://orcid.org/0000-0002-8782-105X"]},"institutions":[{"id":"https://openalex.org/I165799507","display_name":"Rensselaer Polytechnic Institute","ror":"https://ror.org/01rtyzb94","country_code":"US","type":"education","lineage":["https://openalex.org/I165799507"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Kristin P. Bennett","raw_affiliation_strings":["Rensselaer  Polytechnic Institute"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Rensselaer  Polytechnic Institute","institution_ids":["https://openalex.org/I165799507"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5103586972","display_name":"Gregory M. Moore","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I165799507","display_name":"Rensselaer Polytechnic Institute","ror":"https://ror.org/01rtyzb94","country_code":"US","type":"education","lineage":["https://openalex.org/I165799507"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Gregory M. Moore","raw_affiliation_strings":["Rensselaer  Polytechnic Institute"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Rensselaer  Polytechnic Institute","institution_ids":["https://openalex.org/I165799507"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I165799507"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":11,"citation_normalized_percentile":{"value":0.18851094,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":98},"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.6455000042915344,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.6455000042915344,"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/T12535","display_name":"Machine Learning and Data Classification","score":0.31369999051094055,"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/T10100","display_name":"Metaheuristic Optimization Algorithms Research","score":0.004600000102072954,"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/bi-level-optimization","display_name":"bi-level optimization","score":0.9760000109672546},{"id":"https://openalex.org/keywords/hyperparameter-optimization","display_name":"hyperparameter optimization","score":0.925000011920929},{"id":"https://openalex.org/keywords/model-selection","display_name":"model selection","score":0.9589999914169312},{"id":"https://openalex.org/keywords/linear-support-vector-machine","display_name":"linear support vector machine","score":0.7680000066757202},{"id":"https://openalex.org/keywords/nonsmooth-optimization","display_name":"nonsmooth optimization","score":0.7839999794960022},{"id":"https://openalex.org/keywords/mathematical-programs-with-equilibrium-constraints","display_name":"mathematical programs with equilibrium constraints","score":0.6790000200271606},{"id":"https://openalex.org/keywords/nonsmooth-nonconvex-optimization","display_name":"nonsmooth nonconvex optimization","score":0.671999990940094}],"concepts":[{"id":"https://openalex.org/C8642999","wikidata":"https://www.wikidata.org/wiki/Q4171168","display_name":"Hyperparameter","level":2,"score":0.9107120633125305},{"id":"https://openalex.org/C10485038","wikidata":"https://www.wikidata.org/wiki/Q48996162","display_name":"Hyperparameter optimization","level":3,"score":0.8653974533081055},{"id":"https://openalex.org/C3309286","wikidata":"https://www.wikidata.org/wiki/Q4907693","display_name":"Bilevel optimization","level":3,"score":0.7350954413414001},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.6584311723709106},{"id":"https://openalex.org/C12267149","wikidata":"https://www.wikidata.org/wiki/Q282453","display_name":"Support vector machine","level":2,"score":0.6409056186676025},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6142278909683228},{"id":"https://openalex.org/C93959086","wikidata":"https://www.wikidata.org/wiki/Q6888345","display_name":"Model selection","level":2,"score":0.5614822506904602},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.537735104560852},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.4638616442680359},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.4464266002178192},{"id":"https://openalex.org/C81917197","wikidata":"https://www.wikidata.org/wiki/Q628760","display_name":"Selection (genetic algorithm)","level":2,"score":0.42704036831855774},{"id":"https://openalex.org/C137836250","wikidata":"https://www.wikidata.org/wiki/Q984063","display_name":"Optimization problem","level":2,"score":0.350523442029953},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.22979384660720825}],"mesh":[],"locations_count":1,"locations":[{"id":"mag:2287409372","doi":null,"is_oa":false,"landing_page_url":"http://dl.acm.org/citation.cfm?id=2125745","pdf_url":null,"source":null,"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":null}],"best_oa_location":null,"sustainable_development_goals":[],"sustainable_development_goals_aurora":[{"id":"https://metadata.un.org/sdg/9","score":0.6600000262260437,"display_name":"Industry, innovation and infrastructure"}],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W2109346685","https://openalex.org/W2247017865","https://openalex.org/W7097550390","https://openalex.org/W2138218159","https://openalex.org/W1986543815","https://openalex.org/W2608221418","https://openalex.org/W2128255043","https://openalex.org/W7099791564","https://openalex.org/W3110392026","https://openalex.org/W4403690854"],"abstract_inverted_index":{"This":[0,79,525,571,596,640,777],"thesis":[1],"represents":[2],"an":[3,123,144,152,469,560,783],"advance":[4],"towards":[5],"the":[6,44,51,120,131,162,166,174,183,211,215,240,266,273,286,300,314,318,325,337,354,367,384,403,424,440,458,465,475,484,489,507,515,521,533,536,544,552,578,586,609,615,645,650,671,687,692,730,741,763,774],"goal":[7,52],"of":[8,47,53,74,87,93,115,218,257,324,339,452,474,488,535,581,619,649,686,694,735,755,786,802],"self-tuning":[9],"supervised":[10],"data":[11,88,96],"mining":[12],"as":[13,15,136,161,173,222,302,321,387,510],"well":[14,728],"a":[16,70,112,137,303,332,341,377,410,481,527,629,677,702,799,818],"significant":[17],"innovation":[18,400],"in":[19,289,331,359,432,439,577,599,729,753],"scalable":[20,296],"bilevel":[21,26,138,141,178,247,379,445,491,553,662,750],"programming":[22,27,231,248,446],"algorithms.":[23,663,776],"Novel":[24],"nonsmooth":[25,229,234,290,344,411,459,631],"methods":[28,129,193,232,364,421,518,752],"for":[29,182,197,637,670],"training":[30,188,242,523,537,714],"linear":[31,198,235,528],"learning":[32,206],"models":[33,110,220,237],"with":[34,90,147,293,306,353,361,394,409,468,644,681,769,782],"hyperparameters":[35,49,185,270,315,787],"optimized":[36],"via":[37],"T-fold":[38],"cross-validation":[39],"(CV)":[40],"are":[41,58,98,150,170,210,392,497,668,724,794],"considered.":[42],"Determining":[43],"optimal":[45],"values":[46],"these":[48,504],"is":[50,77,143,158,401,454,555,589,605,690],"model":[54,106,132,184,267,319,651,680],"selection.":[55],"Three":[56],"algorithms":[57,243,263,311,329,351,496,505,696,723,743,793],"developed":[59,455,606],"and":[60,67,118,165,187,203,260,269,462,494,558,621,647,658,676,722,733,747,758,796,808],"tested":[61],"against":[62,660],"previous":[63,516,545,622,661],"algorithms:":[64,258],"ImpGrad,":[65],"ImpBundle":[66,495,563,610,659,807],"PBP.":[68,805],"Also":[69],"new":[71,80,103,192,602,678],"generalized":[72],"version":[73],"e-insensitive":[75,673],"regression":[76,674],"developed.":[78],"modeling":[81,756,781],"task,":[82],"called":[83],"multiSVR,":[84],"allows":[85,636],"use":[86,366,520,614],"sets":[89,97],"varying":[91],"levels":[92],"quality.":[94],"Two":[95],"used":[99,423],"to":[100,160,172,213,233,244,250,448,480,502,543,583,607,697,716,726,798],"test":[101],"this":[102,450,626,707],"10-hyperparameter":[104],"model.\r\nCurrent":[105],"selection":[107,133,196,217,652,685],"practice":[108],"constructs":[109],"over":[111],"predefined":[113],"grid":[114,704,745],"hyperparameter":[116,195,216,251],"combinations":[117],"selects":[119,559],"best":[121,764],"one,":[122],"inefficient":[124],"heuristic":[125],"algorithm.":[126,471,611],"The":[127,155,177,191,246,280,327,348,398,719,791],"proposed":[128,208,281,349,374,720,742],"formulate":[130,299],"CV":[134,751],"problem":[135,142,146,301,358,405,467,538,675,689,708],"problem.":[139,176,371],"A":[140,443],"optimization":[145,153,168,226,252,389,512],"constraints":[148,169,308,413],"that":[149,298,365,391,402,436,551,585,635,773],"also":[151,594],"program.":[154],"overall":[156,765],"objective":[157,346],"referred":[159,171],"outer-level":[163],"objective,":[164],"inner":[167],"inner-level":[175,274],"framework":[179],"simultaneously":[180,264],"solves":[181],"(outer-level)":[186],"weights":[189,268,320],"(inner-level).":[190],"address":[194,214],"support":[199],"vector":[200],"machines":[201],"(SVM)":[202],"related":[204],"machine":[205],"problems.\r\nThe":[207],"approaches":[209,249,297],"first":[212,373],"SVM":[219,236,357,370],"expressed":[221],"unconstrained":[223,356,388,511],"(potentially":[224],"nonsmooth)":[225],"problems.":[227,820],"Applying":[228],"mathematical":[230,304],"has":[238],"produced":[239],"fastest":[241],"date.":[245],"can":[253,406,539],"produce":[254],"two":[255],"types":[256],"explicit":[259,282,375],"implicit.":[261],"Explicit":[262],"optimize":[265,312],"by":[271,276,316,456,567],"replacing":[272],"problems":[275,386,390,453,509,667,715,803],"their":[277,395],"optimality":[278,287,396,434,460,486,529],"conditions.":[279,397],"algorithm":[283,447,476,573,604,641],"PBP":[284,478,761],"utilizes":[285],"conditions":[288,435,487],"form,":[291],"contrasting":[292],"prior":[294,362,748],"less":[295,813],"program":[305,380,554],"equilibrium":[307],"(MPEC).":[309],"Implicit":[310],"only":[313,422,591],"treating":[317],"implicit":[322,328,350,363,498,517,546,623,721,775,792],"functions":[323],"hyperparameters.":[326,683,736],"work":[330,352],"low":[333],"dimensional":[334],"space":[335],"at":[336],"expense":[338],"having":[340],"very":[342],"challenging":[343],"nonconvex":[345,632],"function.":[347],"primal":[355,426],"contrast":[360],"smooth":[368,749],"dual":[369,428,522],"\r\nThe":[372],"approach,":[376],"penalized":[378,490],"(PBP)":[381],"algorithm,":[382,548],"treats":[383],"lower-level":[385,508],"replaced":[393,408],"key":[399],"innerlevel":[404],"be":[407,540,717],"nonlinear":[412,433],"without":[414],"introducing":[415],"extra":[416],"variables":[417],"or":[418,427],"constraints.":[419,639],"Previous":[420],"smoothed":[425],"formulations":[429],"which":[430,519,532],"results":[431,654,738,767,816],"grow":[437],"exponentially":[438],"sample":[441,731],"size.":[442],"novel":[444,630],"solve":[449,608,698],"class":[451,801],"penalizing":[457],"condition":[461,530],"then":[463],"solving":[464],"resulting":[466],"approximation":[470],"Convergence":[472],"analysis":[473],"shows":[477],"converges":[479],"solution":[482],"satisfying":[483],"necessary":[485],"program.\r\nThe":[492],"ImpGrad":[493,549,566,620,657,810],"gradient-based":[499,547,624],"approaches.":[500],"Similar":[501,542],"PBP,":[503,656],"treat":[506],"problems,":[513],"unlike":[514],"formulation.":[524],"yields":[526],"from":[531],"gradient":[534],"computed.":[541],"assumes":[550],"locally":[556],"smooth,":[557],"arbitrary":[561],"subgradient.":[562],"expands":[564],"on":[565,665,706,788,817],"not":[568,590],"assuming":[569],"smoothness.":[570],"improved":[572],"uses":[574,628],"directional":[575],"derivatives":[576],"search":[579,705,746],"direction":[580],"interest":[582],"ensure":[584],"subgradient":[587],"chosen":[588],"valid,":[592],"but":[593,768],"informative.":[595],"prevents":[597],"searching":[598],"assent":[600],"directions.\r\nA":[601],"outer":[603],"Rather":[612],"than":[613,740,804],"Broyden-Fletcher-Goldfarb-Shanno":[616],"(BFGS)":[617],"method":[618,627,634],"approaches,":[625],"bundle":[633],"bounds":[638],"explicitly":[642],"deals":[643],"nonsmoothness":[646],"nonconvexity":[648],"problem.\r\nComputational":[653],"compare":[655],"Results":[664],"cheminformatics":[666],"presented":[669],"traditional":[672],"multiSVR":[679],"10":[682],"Model":[684],"later":[688],"beyond":[691],"ability":[693],"most":[695],"efficiently.":[699],"In":[700],"fact":[701],"thorough":[703],"would":[709],"require":[710],"roughly":[711],"250":[712],"million":[713],"solved.":[718],"shown":[725],"scale":[727],"size":[732],"number":[734,785],"Experimental":[737],"show":[739],"outperform":[744],"terms":[754],"performance":[757],"computational":[759,771],"time.":[760],"produces":[762],"generalization":[766,815],"larger":[770,800],"times":[772],"increased":[778,784],"speed":[779],"permits":[780],"massive":[789],"datasets.":[790],"faster":[795],"applicable":[797],"But":[806],"particularly":[809],"yielded":[811],"slightly":[812],"robust":[814],"few":[819]},"counts_by_year":[{"year":2023,"cited_by_count":1},{"year":2022,"cited_by_count":2},{"year":2021,"cited_by_count":5},{"year":2020,"cited_by_count":2},{"year":2016,"cited_by_count":1}],"updated_date":"2026-10-12T09:40:04.860316","created_date":"2016-06-24T00:00:00"}