{"id":"https://openalex.org/W3015208560","doi":null,"title":"Large-Margin Structured Prediction Extensions of Neural Networks for Automatic Speech Recognition","display_name":"Large-Margin Structured Prediction Extensions of Neural Networks for Automatic Speech Recognition","publication_year":2015,"publication_date":"2015-01-01","ids":{"openalex":"https://openalex.org/W3015208560","mag":"3015208560"},"language":"en","primary_location":{"id":"pmh:oai:berkeley.edu:28722-bk0016x908s","doi":null,"is_oa":false,"landing_page_url":"http://techreports.lib.berkeley.edu/reports/TRWebData/accessPages/EECS-2015-241.html","pdf_url":null,"source":{"id":"https://openalex.org/S7407050879","display_name":"UC Berkeley","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I95457486","host_organization_name":"University of California, Berkeley","host_organization_lineage":["https://openalex.org/I95457486"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"technical reports"},"volume":null,"type":"report","is_traditional":false,"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/A5026262869","display_name":"Suman V. Ravuri","orcid":"https://orcid.org/0000-0002-7481-7633","observed_orcids":["https://orcid.org/0000-0002-7481-7633"]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Suman Ravuri","raw_affiliation_strings":[],"raw_orcid":"https://orcid.org/0000-0002-7481-7633","affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5026262869"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"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/T10201","display_name":"Speech Recognition and Synthesis","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/T10201","display_name":"Speech Recognition and Synthesis","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/T10860","display_name":"Speech and Audio Processing","score":0.0,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T12031","display_name":"Speech and dialogue systems","score":0.0,"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"}}],"keywords":[{"id":"https://openalex.org/keywords/structured-support-vector-machines","display_name":"structured support vector machines","score":0.8040000200271606},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"speech recognition","score":0.8159999847412109},{"id":"https://openalex.org/keywords/deep-neural-network","display_name":"deep neural network","score":0.8289999961853027},{"id":"https://openalex.org/keywords/hidden-markov-models","display_name":"hidden Markov models","score":0.5590000152587891},{"id":"https://openalex.org/keywords/conditional-independence-assumption","display_name":"conditional independence assumption","score":0.6499999761581421},{"id":"https://openalex.org/keywords/bootstrap-method","display_name":"bootstrapping","score":0.7459999918937683},{"id":"https://openalex.org/keywords/log-linear-models","display_name":"log-linear models","score":0.6190000176429749},{"id":"https://openalex.org/keywords/maximum-mutual-information","display_name":"maximum mutual information","score":0.7129999995231628},{"id":"https://openalex.org/keywords/minimum-phone-error","display_name":"minimum phone error","score":0.8230000138282776},{"id":"https://openalex.org/keywords/minimum-bayes-risk","display_name":"minimum Bayes risk","score":0.9580000042915344},{"id":"https://openalex.org/keywords/online-training","display_name":"online training","score":0.47999998927116394},{"id":"https://openalex.org/keywords/word-error-rate","display_name":"word error rate","score":0.6449999809265137},{"id":"https://openalex.org/keywords/feature-construction","display_name":"feature construction","score":0.4749999940395355},{"id":"https://openalex.org/keywords/hinge-loss","display_name":"hinge loss","score":0.3880000114440918}],"concepts":[{"id":"https://openalex.org/C23224414","wikidata":"https://www.wikidata.org/wiki/Q176769","display_name":"Hidden Markov model","level":2,"score":0.7621654272079468},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7218096256256104},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.6461209654808044},{"id":"https://openalex.org/C774472","wikidata":"https://www.wikidata.org/wiki/Q6760393","display_name":"Margin (machine learning)","level":2,"score":0.5783281326293945},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.5771788358688354},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5590563416481018},{"id":"https://openalex.org/C169903167","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Test set","level":2,"score":0.5111353993415833},{"id":"https://openalex.org/C12267149","wikidata":"https://www.wikidata.org/wiki/Q282453","display_name":"Support vector machine","level":2,"score":0.5103477835655212},{"id":"https://openalex.org/C79772020","wikidata":"https://www.wikidata.org/wiki/Q5159264","display_name":"Conditional independence","level":2,"score":0.49321144819259644},{"id":"https://openalex.org/C61224824","wikidata":"https://www.wikidata.org/wiki/Q2260434","display_name":"Mixture model","level":2,"score":0.46835651993751526},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.4680373966693878},{"id":"https://openalex.org/C150921843","wikidata":"https://www.wikidata.org/wiki/Q1170431","display_name":"Resampling","level":2,"score":0.4667143225669861},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.4476274847984314},{"id":"https://openalex.org/C2776401178","wikidata":"https://www.wikidata.org/wiki/Q12050496","display_name":"Feature (linguistics)","level":2,"score":0.44467097520828247},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.40763503313064575},{"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/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"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":5,"locations":[{"id":"pmh:oai:berkeley.edu:28722-bk0016x908s","doi":null,"is_oa":false,"landing_page_url":"http://techreports.lib.berkeley.edu/reports/TRWebData/accessPages/EECS-2015-241.html","pdf_url":null,"source":{"id":"https://openalex.org/S7407050879","display_name":"UC Berkeley","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I95457486","host_organization_name":"University of California, Berkeley","host_organization_lineage":["https://openalex.org/I95457486"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"technical reports"},{"id":"pmh:oai:berkeley.edu:28722-bk001807m56","doi":null,"is_oa":false,"landing_page_url":"http://techreports.lib.berkeley.edu/reports/TRWebData/accessPages/EECS-2017-169.html","pdf_url":null,"source":{"id":"https://openalex.org/S7407050879","display_name":"UC Berkeley","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I95457486","host_organization_name":"University of California, Berkeley","host_organization_lineage":["https://openalex.org/I95457486"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"technical reports"},{"id":"pmh:oai:escholarship.org/ark:/13030/qt5fk9p6vw","doi":null,"is_oa":false,"landing_page_url":"https://escholarship.org/uc/item/5fk9p6vw","pdf_url":null,"source":{"id":"https://openalex.org/S4306400115","display_name":"eScholarship (California Digital Library)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I2801248553","host_organization_name":"California Digital Library","host_organization_lineage":["https://openalex.org/I2801248553"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"etd"},{"id":"pmh:qt5fk9p6vw","doi":null,"is_oa":false,"landing_page_url":"http://www.escholarship.org/uc/item/5fk9p6vw","pdf_url":null,"source":{"id":"https://openalex.org/S4306400115","display_name":"eScholarship (California Digital Library)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I2801248553","host_organization_name":"California Digital Library","host_organization_lineage":["https://openalex.org/I2801248553"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"Ravuri, Suman. (2015). Large-Margin Structured Prediction Extensions of Neural Networks for Automatic Speech Recognition. UC Berkeley: Electrical Engineering & Computer Sciences. Retrieved from: http://www.escholarship.org/uc/item/5fk9p6vw","raw_type":"dissertation"},{"id":"mag:3015208560","doi":null,"is_oa":false,"landing_page_url":"https://escholarship.org/content/qt5fk9p6vw/qt5fk9p6vw.pdf?t=p6pil7","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":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":38,"referenced_works":["https://openalex.org/W38527073","https://openalex.org/W75470185","https://openalex.org/W196761320","https://openalex.org/W811578723","https://openalex.org/W1513984940","https://openalex.org/W1524333225","https://openalex.org/W1559401186","https://openalex.org/W1877570817","https://openalex.org/W1922655562","https://openalex.org/W1932968309","https://openalex.org/W1967573895","https://openalex.org/W1978741356","https://openalex.org/W1993666847","https://openalex.org/W2008652694","https://openalex.org/W2048608953","https://openalex.org/W2053567709","https://openalex.org/W2083751884","https://openalex.org/W2099119623","https://openalex.org/W2117722866","https://openalex.org/W2117897510","https://openalex.org/W2121573996","https://openalex.org/W2124181495","https://openalex.org/W2125993116","https://openalex.org/W2139686264","https://openalex.org/W2147511255","https://openalex.org/W2147768505","https://openalex.org/W2147880316","https://openalex.org/W2150907703","https://openalex.org/W2156373089","https://openalex.org/W2158195707","https://openalex.org/W2161329381","https://openalex.org/W2165712214","https://openalex.org/W2171631590","https://openalex.org/W2277634955","https://openalex.org/W2290722036","https://openalex.org/W2295676751","https://openalex.org/W2403195671","https://openalex.org/W2405662050"],"related_works":["https://openalex.org/W2290722036","https://openalex.org/W2159055159","https://openalex.org/W2160815625","https://openalex.org/W1794977688","https://openalex.org/W2600249726","https://openalex.org/W124924985","https://openalex.org/W2063689849","https://openalex.org/W7099721616","https://openalex.org/W7043080823","https://openalex.org/W2400718301"],"abstract_inverted_index":{"Neural":[0,36],"networks,":[1],"especially":[2],"those":[3],"with":[4,174,218],"more":[5,187],"than":[6,400],"one":[7,137,208],"hidden":[8,112,162,220],"layer,":[9,155],"have":[10],"re-emerged":[11],"in":[12,46,121],"Automatic":[13],"Speech":[14],"Recognition":[15],"(ASR)":[16],"systems":[17,127],"as":[18,164,258],"replacements":[19],"to":[20,44,132,166,186,192,264,276,287,295,301],"emission":[21],"models":[22,394],"based":[23],"on":[24,128,200],"Gaussian":[25],"Mixture":[26],"Models":[27],"(GMMs).":[28],"While":[29],"the":[30,49,76,111,118,144,153,291,307,320,324,333,348,372,377,405,412],"use":[31,334],"of":[32,52,75,110,160,248,255,319,335,347],"these":[33,116],"so-called":[34],"Deep":[35],"Networks":[37],"(DNNs)":[38],"has":[39],"enjoyed":[40],"widespread":[41],"success":[42],"due":[43],"improvements":[45],"recognition":[47,54,140],"results,":[48],"exact":[50],"source":[51],"better":[53,211,341,396],"accuracy":[55],"is":[56,315],"not":[57],"entirely":[58],"understood.":[59],"Using":[60],"a":[61,167,215,219,229,253,330],"bootstrap":[62,378],"resampling":[63,379],"framework":[64],"that":[65,85,136,172,240],"generates":[66],"synthetic":[67,133,401],"test":[68,201],"set":[69,202],"data":[70,130,134],"satisfying":[71],"conditional":[72,100],"independence":[73,101],"assumptions":[74,102,410],"model":[77,173,234,250,272,406],"while":[78,282],"still":[79],"using":[80,158,376],"real":[81,129,398],"observations,":[82],"I":[83,156,205,227,289,304,344],"show":[84,206,305],"DNNs":[86],"used":[87,263],"for":[88,98,125,244,310,323,340,389,397],"both":[89],"feature":[90],"generation":[91],"and":[92,103,170,196,235,298,327,365,371,381],"hybrid":[93,230],"acoustic":[94,233],"modeling":[95],"help":[96,104],"compensate":[97,388],"incorrect":[99],"fix":[105],"poor":[106],"phone":[107],"duration":[108],"estimates":[109],"Markov":[113,221],"Model":[114],"(HMM).Despite":[115],"improvements,":[117],"large":[119],"increase":[120],"word":[122],"error":[123],"rates":[124],"DNN-HMM":[126],"compared":[131],"suggests":[135],"can":[138,209],"improve":[139],"performance":[141],"by":[142,213,279],"modifying":[143],"training":[145,171,238,278,297,352,386],"criterion.":[146],"Since":[147],"neural":[148,216],"networks":[149],"are":[150,262],"log-linear":[151,168],"at":[152],"output":[154],"propose":[157,228],"sequences":[159],"last":[161],"layers":[163],"input":[165],"model,":[169],"large-margin":[175],"criteria.":[176],"These":[177],"Structured":[178,270,292,325,374,392],"Support":[179,222],"Vector":[180,223],"Machine":[181,224],"(SVM)":[182],"approaches":[183,256],"allow":[184],"us":[185],"directly":[188],"minimize":[189],"errors":[190],"relevant":[191],"automatic":[193],"speech":[194],"recognition,":[195],"provide":[197],"some":[198],"guarantees":[199],"error.":[203],"First,":[204],"how":[207,306,328,383],"generate":[210],"features":[212],"combining":[214],"network":[217],"(HMSVM).":[225],"Then,":[226],"DNN-Structured":[231],"SVM":[232,271,293,375,393],"an":[236,316],"online":[237],"algorithm":[239],"iteratively":[241],"updates":[242],"alignments":[243],"faster":[245],"convergence.":[246],"Training":[247],"this":[249],"falls":[251],"under":[252],"class":[254],"known":[257],"sequence-discriminative":[259,277,296,351,385],"training,":[260],"which":[261],"train":[265],"state-of-the-art":[266],"systems.":[267],"This":[268],"DNN-latent":[269],"beats":[273],"alternative":[274],"methods":[275],"1.0%":[280],"absolute,":[281],"needing":[283],"33-66%":[284],"fewer":[285,408],"utterances":[286],"converge.Finally,":[288],"analyze":[290,345],"approach":[294],"compare":[299,382],"it":[300],"standard":[302],"methods.":[303],"loss":[308,322],"function":[309],"boosted":[311,358],"Maximum":[312,355,359],"Mutual":[313,356,360],"Information":[314],"upper":[317],"bound":[318],"hinge":[321],"SVM,":[326],"such":[329],"relaxation":[331],"precludes":[332],"aggressive":[336],"boosting":[337],"parameters":[338],"needed":[339],"results.":[342],"Finally,":[343],"four":[346],"most":[349],"popular":[350],"criteria":[353,387],"\u2013":[354,370],"Information,":[357,361],"Minimum":[362,367],"Phone":[363],"Error,":[364],"state-level":[366],"Bayes":[368],"Risk":[369],"latent":[373],"framework,":[380],"different":[384],"data/model":[390],"mismatch.":[391],"perform":[395],"rather":[399],"data,":[402],"likely":[403],"because":[404],"makes":[407],"distributional":[409],"about":[411],"underlying":[413],"data.":[414]},"counts_by_year":[],"updated_date":"2026-11-11T12:16:26.678958","created_date":"2020-04-17T00:00:00"}