{"id":"https://openalex.org/W3095229326","doi":"https://doi.org/10.21437/interspeech.2020-1569","title":"Subword Regularization: An Analysis of Scalability and Generalization for End-to-End Automatic Speech Recognition","display_name":"Subword Regularization: An Analysis of Scalability and Generalization for End-to-End Automatic Speech Recognition","publication_year":2020,"publication_date":"2020-10-25","ids":{"openalex":"https://openalex.org/W3095229326","doi":"https://doi.org/10.21437/interspeech.2020-1569","mag":"3095229326"},"language":"en","primary_location":{"is_oa":false,"landing_page_url":"https://doi.org/10.21437/interspeech.2020-1569","pdf_url":null,"source":{"id":"https://openalex.org/S4363604309","display_name":"Interspeech 2022","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false},"type":"article","type_crossref":"proceedings-article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"http://arxiv.org/pdf/2008.04034","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5045428440","display_name":"Egor Lakomkin","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Egor Lakomkin","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5006131416","display_name":"Jahn Heymann","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Jahn Heymann","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5064043969","display_name":"Ilya Sklyar","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ilya Sklyar","raw_affiliation_strings":[],"affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5077727156","display_name":"Simon Wiesler","orcid":null},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Simon Wiesler","raw_affiliation_strings":[],"affiliations":[]}],"institution_assertions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.73,"has_fulltext":false,"cited_by_count":9,"citation_normalized_percentile":{"value":0.632479,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":85,"max":86},"biblio":{"volume":null,"issue":null,"first_page":"3600","last_page":"3604"},"is_retracted":false,"is_paratext":false,"primary_topic":{"id":"https://openalex.org/T10201","display_name":"Speech Recognition and Synthesis","score":0.9988,"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":0.9988,"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":0.9941,"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/T10028","display_name":"Topic Modeling","score":0.9757,"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/regularization","display_name":"Regularization","score":0.65866774},{"id":"https://openalex.org/keywords/end-to-end-principle","display_name":"End-to-end principle","score":0.64205253}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.71806353},{"id":"https://openalex.org/C2776135515","wikidata":"https://www.wikidata.org/wiki/Q17143721","display_name":"Regularization (linguistics)","level":2,"score":0.65866774},{"id":"https://openalex.org/C74296488","wikidata":"https://www.wikidata.org/wiki/Q2527392","display_name":"End-to-end principle","level":2,"score":0.64205253},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.62150615},{"id":"https://openalex.org/C48044578","wikidata":"https://www.wikidata.org/wiki/Q727490","display_name":"Scalability","level":2,"score":0.5354289},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.50900817},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.40119767},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.36322945},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.11494872},{"id":"https://openalex.org/C77088390","wikidata":"https://www.wikidata.org/wiki/Q8513","display_name":"Database","level":1,"score":0.05860856},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"is_oa":false,"landing_page_url":"https://doi.org/10.21437/interspeech.2020-1569","pdf_url":null,"source":{"id":"https://openalex.org/S4363604309","display_name":"Interspeech 2022","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false},{"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2008.04034","pdf_url":"http://arxiv.org/pdf/2008.04034","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":["Cornell University"],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false}],"best_oa_location":{"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2008.04034","pdf_url":"http://arxiv.org/pdf/2008.04034","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":["Cornell University"],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/4","display_name":"Quality education","score":0.69}],"grants":[],"datasets":[],"versions":[],"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W3087771547","https://openalex.org/W2390485179","https://openalex.org/W2382623646","https://openalex.org/W2375199418","https://openalex.org/W2368437561","https://openalex.org/W2364921833","https://openalex.org/W2333420780","https://openalex.org/W2302028273","https://openalex.org/W2067938758","https://openalex.org/W1525643724"],"abstract_inverted_index":{"Subwords":[0],"are":[1,63,68],"the":[2,14,21,29,72,110,114,130,136,139,174],"most":[3,73],"widely":[4],"used":[5],"output":[6],"units":[7,42],"in":[8,158],"end-to-end":[9,122],"speech":[10,35,94,123],"recognition.":[11],"They":[12],"combine":[13],"best":[15],"of":[16,23,113,138,152,168,176,181],"two":[17],"worlds":[18],"by":[19,37],"modeling":[20],"majority":[22],"frequent":[24],"words":[25,47,183],"directly":[26],"and":[27,58,93,184],"at":[28],"same":[30],"time":[31],"allow":[32],"open":[33],"vocabulary":[34],"recognition":[36,95,124,180],"backing":[38],"off":[39],"to":[40,45,54,98,165],"shorter":[41],"or":[43],"characters":[44],"construct":[46],"unseen":[48,182],"during":[49,83],"training.":[50],"However,":[51],"mapping":[52],"text":[53],"subwords":[55],"is":[56],"ambiguous":[57],"often":[59],"multiple":[60],"segmentation":[61,116],"variants":[62],"possible.":[64],"Yet,":[65],"many":[66],"systems":[67],"trained":[69],"using":[70],"only":[71],"likely":[74],"segmentation.":[75],"Recent":[76],"research":[77],"suggests":[78],"that":[79,145],"sampling":[80,117],"subword":[81,115,131,146,177],"segmentations":[82],"training":[84,140],"acts":[85],"as":[86],"a":[87,106,120,149,159,166],"regularizer":[88],"for":[89,119],"neural":[90],"machine":[91],"translation":[92],"models,":[96],"leading":[97],"performance":[99],"improvements.":[100],"In":[101,126],"this":[102],"work,":[103],"we":[104,128,172],"conduct":[105],"principled":[107],"investigation":[108],"on":[109,135,179,187],"regularizing":[111],"effect":[112,175],"method":[118],"streaming":[121],"task.":[125],"particular,":[127],"evaluate":[129],"regularization":[132,147,178],"contribution":[133],"depending":[134],"size":[137,167],"dataset.":[141],"Our":[142],"results":[143],"suggest":[144],"provides":[148],"consistent":[150],"improvement":[151],"(2-8%)":[153],"relative":[154],"word-error-rate":[155],"reduction,":[156],"even":[157],"large-scale":[160],"setting":[161],"with":[162],"datasets":[163],"up":[164],"20k":[169],"hours.":[170],"Further,":[171],"analyze":[173],"its":[185],"implications":[186],"beam":[188],"diversity.":[189]},"cited_by_api_url":"https://api.openalex.org/works?filter=cites:W3095229326","counts_by_year":[{"year":2023,"cited_by_count":2},{"year":2022,"cited_by_count":2},{"year":2021,"cited_by_count":5}],"updated_date":"2024-12-07T13:59:44.280389","created_date":"2020-11-09"}