{"id":"https://openalex.org/W2997454826","doi":"https://doi.org/10.1109/asru46091.2019.9003936","title":"Improved Multi-Stage Training of Online Attention-Based Encoder-Decoder Models","display_name":"Improved Multi-Stage Training of Online Attention-Based Encoder-Decoder Models","publication_year":2019,"publication_date":"2019-12-01","ids":{"openalex":"https://openalex.org/W2997454826","doi":"https://doi.org/10.1109/asru46091.2019.9003936","mag":"2997454826"},"language":"en","primary_location":{"id":"doi:10.1109/asru46091.2019.9003936","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asru46091.2019.9003936","pdf_url":null,"source":{"id":"https://openalex.org/S4306498489","display_name":"2019 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2019 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)","raw_type":"proceedings-article"},"type":"article","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/A5043920881","display_name":"Abhinav Garg","orcid":"https://orcid.org/0000-0001-5082-5500"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":true,"raw_author_name":"Abhinav Garg","raw_affiliation_strings":["Speech Processing Lab, AI Center, Samsung Research, Korea"],"affiliations":[{"raw_affiliation_string":"Speech Processing Lab, AI Center, Samsung Research, Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5113125574","display_name":"Dhananjaya Gowda","orcid":null},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Dhananjaya Gowda","raw_affiliation_strings":["Speech Processing Lab, AI Center, Samsung Research, Korea"],"affiliations":[{"raw_affiliation_string":"Speech Processing Lab, AI Center, Samsung Research, Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5053775185","display_name":"Ankur Kumar","orcid":"https://orcid.org/0000-0002-9248-1416"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Ankur Kumar","raw_affiliation_strings":["Speech Processing Lab, AI Center, Samsung Research, Korea"],"affiliations":[{"raw_affiliation_string":"Speech Processing Lab, AI Center, Samsung Research, Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5052377938","display_name":"Kwangyoun Kim","orcid":null},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Kwangyoun Kim","raw_affiliation_strings":["Speech Processing Lab, AI Center, Samsung Research, Korea"],"affiliations":[{"raw_affiliation_string":"Speech Processing Lab, AI Center, Samsung Research, Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5065995328","display_name":"Mehul Kumar","orcid":"https://orcid.org/0000-0001-7341-2362"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Mehul Kumar","raw_affiliation_strings":["Speech Processing Lab, AI Center, Samsung Research, Korea"],"affiliations":[{"raw_affiliation_string":"Speech Processing Lab, AI Center, Samsung Research, Korea","institution_ids":["https://openalex.org/I2250650973"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100684422","display_name":"Chanwoo Kim","orcid":"https://orcid.org/0000-0003-0193-8167"},"institutions":[{"id":"https://openalex.org/I2250650973","display_name":"Samsung (South Korea)","ror":"https://ror.org/04w3jy968","country_code":"KR","type":"company","lineage":["https://openalex.org/I2250650973"]}],"countries":["KR"],"is_corresponding":false,"raw_author_name":"Chanwoo Kim","raw_affiliation_strings":["Speech Processing Lab, AI Center, Samsung Research, Korea"],"affiliations":[{"raw_affiliation_string":"Speech Processing Lab, AI Center, Samsung Research, Korea","institution_ids":["https://openalex.org/I2250650973"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5043920881"],"corresponding_institution_ids":["https://openalex.org/I2250650973"],"apc_list":null,"apc_paid":null,"fwci":2.4568,"has_fulltext":false,"cited_by_count":19,"citation_normalized_percentile":{"value":0.91878721,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":90,"max":98},"biblio":{"volume":"abs 1801 59","issue":null,"first_page":"70","last_page":"77"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","score":0.9998999834060669,"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/T10028","display_name":"Topic Modeling","score":0.9998999834060669,"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/T10181","display_name":"Natural Language Processing Techniques","score":0.9998000264167786,"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/T10201","display_name":"Speech Recognition and Synthesis","score":0.9990000128746033,"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/encoder","display_name":"Encoder","score":0.8704309463500977},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.7940977811813354},{"id":"https://openalex.org/keywords/task","display_name":"Task (project management)","score":0.6462895274162292},{"id":"https://openalex.org/keywords/byte","display_name":"Byte","score":0.6432778835296631},{"id":"https://openalex.org/keywords/encoding","display_name":"Encoding (memory)","score":0.5593403577804565},{"id":"https://openalex.org/keywords/character","display_name":"Character (mathematics)","score":0.5446876287460327},{"id":"https://openalex.org/keywords/granularity","display_name":"Granularity","score":0.5262053608894348},{"id":"https://openalex.org/keywords/language-model","display_name":"Language model","score":0.5186483263969421},{"id":"https://openalex.org/keywords/word-error-rate","display_name":"Word error rate","score":0.5057594776153564},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4916718602180481},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.4671209454536438},{"id":"https://openalex.org/keywords/decoding-methods","display_name":"Decoding methods","score":0.458281934261322},{"id":"https://openalex.org/keywords/code","display_name":"Code (set theory)","score":0.4412826597690582},{"id":"https://openalex.org/keywords/word","display_name":"Word (group theory)","score":0.42344918847084045},{"id":"https://openalex.org/keywords/task-analysis","display_name":"Task analysis","score":0.414425253868103},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.36350613832473755},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.22101768851280212},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.16869962215423584},{"id":"https://openalex.org/keywords/programming-language","display_name":"Programming language","score":0.09519258141517639},{"id":"https://openalex.org/keywords/linguistics","display_name":"Linguistics","score":0.07593747973442078},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.07279518246650696}],"concepts":[{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.8704309463500977},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7940977811813354},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.6462895274162292},{"id":"https://openalex.org/C43364308","wikidata":"https://www.wikidata.org/wiki/Q8799","display_name":"Byte","level":2,"score":0.6432778835296631},{"id":"https://openalex.org/C125411270","wikidata":"https://www.wikidata.org/wiki/Q18653","display_name":"Encoding (memory)","level":2,"score":0.5593403577804565},{"id":"https://openalex.org/C2780861071","wikidata":"https://www.wikidata.org/wiki/Q1062934","display_name":"Character (mathematics)","level":2,"score":0.5446876287460327},{"id":"https://openalex.org/C177774035","wikidata":"https://www.wikidata.org/wiki/Q1246948","display_name":"Granularity","level":2,"score":0.5262053608894348},{"id":"https://openalex.org/C137293760","wikidata":"https://www.wikidata.org/wiki/Q3621696","display_name":"Language model","level":2,"score":0.5186483263969421},{"id":"https://openalex.org/C40969351","wikidata":"https://www.wikidata.org/wiki/Q3516228","display_name":"Word error rate","level":2,"score":0.5057594776153564},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4916718602180481},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.4671209454536438},{"id":"https://openalex.org/C57273362","wikidata":"https://www.wikidata.org/wiki/Q576722","display_name":"Decoding methods","level":2,"score":0.458281934261322},{"id":"https://openalex.org/C2776760102","wikidata":"https://www.wikidata.org/wiki/Q5139990","display_name":"Code (set theory)","level":3,"score":0.4412826597690582},{"id":"https://openalex.org/C90805587","wikidata":"https://www.wikidata.org/wiki/Q10944557","display_name":"Word (group theory)","level":2,"score":0.42344918847084045},{"id":"https://openalex.org/C175154964","wikidata":"https://www.wikidata.org/wiki/Q380077","display_name":"Task analysis","level":3,"score":0.414425253868103},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.36350613832473755},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.22101768851280212},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.16869962215423584},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.09519258141517639},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.07593747973442078},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.07279518246650696},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.0},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.0},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0},{"id":"https://openalex.org/C177264268","wikidata":"https://www.wikidata.org/wiki/Q1514741","display_name":"Set (abstract data type)","level":2,"score":0.0},{"id":"https://openalex.org/C187736073","wikidata":"https://www.wikidata.org/wiki/Q2920921","display_name":"Management","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":1,"locations":[{"id":"doi:10.1109/asru46091.2019.9003936","is_oa":false,"landing_page_url":"https://doi.org/10.1109/asru46091.2019.9003936","pdf_url":null,"source":{"id":"https://openalex.org/S4306498489","display_name":"2019 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2019 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"score":0.4300000071525574,"id":"https://metadata.un.org/sdg/9","display_name":"Industry, innovation and infrastructure"}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":50,"referenced_works":["https://openalex.org/W1494198834","https://openalex.org/W1522301498","https://openalex.org/W1524333225","https://openalex.org/W1533416326","https://openalex.org/W1836465849","https://openalex.org/W1915251500","https://openalex.org/W2095705004","https://openalex.org/W2127141656","https://openalex.org/W2191779130","https://openalex.org/W2327501763","https://openalex.org/W2402144811","https://openalex.org/W2514741789","https://openalex.org/W2526425061","https://openalex.org/W2555428947","https://openalex.org/W2597757402","https://openalex.org/W2605141709","https://openalex.org/W2606621590","https://openalex.org/W2627092829","https://openalex.org/W2747874407","https://openalex.org/W2750499125","https://openalex.org/W2767487732","https://openalex.org/W2782451907","https://openalex.org/W2799800213","https://openalex.org/W2799923439","https://openalex.org/W2888884590","https://openalex.org/W2901837390","https://openalex.org/W2932319281","https://openalex.org/W2936774411","https://openalex.org/W2949117887","https://openalex.org/W2953384591","https://openalex.org/W2962728618","https://openalex.org/W2962760690","https://openalex.org/W2962784628","https://openalex.org/W2962824709","https://openalex.org/W2963070863","https://openalex.org/W2964038834","https://openalex.org/W2964121744","https://openalex.org/W2964272710","https://openalex.org/W2973215447","https://openalex.org/W3105532142","https://openalex.org/W4294619417","https://openalex.org/W6631190155","https://openalex.org/W6631362777","https://openalex.org/W6638667902","https://openalex.org/W6640059789","https://openalex.org/W6713134421","https://openalex.org/W6735706088","https://openalex.org/W6736722209","https://openalex.org/W6747158283","https://openalex.org/W6747836893"],"related_works":["https://openalex.org/W2594897229","https://openalex.org/W2151348424","https://openalex.org/W4221142855","https://openalex.org/W2050138804","https://openalex.org/W4290708361","https://openalex.org/W2129812225","https://openalex.org/W26527944","https://openalex.org/W2523799048","https://openalex.org/W2155620340","https://openalex.org/W2944691285"],"abstract_inverted_index":{"In":[0],"this":[1],"paper,":[2],"we":[3],"propose":[4],"a":[5,75,102],"refined":[6],"multi-stage":[7],"multi-task":[8,49],"training":[9,26],"strategy":[10],"to":[11],"improve":[12],"the":[13,69,112,117],"performance":[14],"of":[15,31,54,107],"onli":[16],"<sup":[17],"xmlns:mml=\"http://www.w3.org/1998/Math/MathML\"":[18],"xmlns:xlink=\"http://www.w3.org/1999/xlink\">ne</sup>":[19],"attention-based":[20],"encoder-decoder":[21,79],"(AED)":[22],"models.":[23],"A":[24],"three-stage":[25],"based":[27,41,51,130],"on":[28,52,111],"three":[29],"levels":[30],"architectural":[32],"granularity":[33,56],"namely,":[34,57],"character":[35,58],"encoder,":[36,42],"byte":[37],"pair":[38],"encoding":[39],"(BPE)":[40],"and":[43,59,86,95,109,119],"attention":[44,83],"decoder,":[45],"is":[46,61],"proposed.":[47],"Also,":[48],"learning":[50,73],"two-levels":[53],"linguistic":[55],"BPE,":[60],"used.":[62],"We":[63],"explore":[64],"different":[65],"pre-training":[66],"strategies":[67],"for":[68,93,116],"encoders":[70],"including":[71],"transfer":[72],"from":[74],"bidirectional":[76],"encoder.":[77],"Our":[78,99],"models":[80,100,121],"with":[81,125],"online":[82],"show":[84],"~35%":[85],"~10%":[87],"relative":[88],"improvement":[89],"over":[90],"their":[91],"baselines":[92],"smaller":[94,118],"bigger":[96,120],"models,":[97],"respectively.":[98],"achieve":[101],"word":[103],"error":[104],"rate":[105],"(WER)":[106],"5.04%":[108],"4.48%":[110],"Librispeech":[113],"test-clean":[114],"data":[115],"respectively":[122],"after":[123],"fusion":[124],"long":[126],"short-term":[127],"memory":[128],"(LSTM)":[129],"external":[131],"language":[132],"model":[133],"(LM).":[134]},"counts_by_year":[{"year":2024,"cited_by_count":1},{"year":2023,"cited_by_count":1},{"year":2022,"cited_by_count":2},{"year":2021,"cited_by_count":7},{"year":2020,"cited_by_count":4},{"year":2019,"cited_by_count":4}],"updated_date":"2026-04-17T18:11:37.981687","created_date":"2025-10-10T00:00:00"}