{"id":"https://openalex.org/W4297206661","doi":"https://doi.org/10.48550/arxiv.2204.06260","title":"Self-critical Sequence Training for Automatic Speech Recognition","display_name":"Self-critical Sequence Training for Automatic Speech Recognition","publication_year":2022,"publication_date":"2022-04-13","ids":{"openalex":"https://openalex.org/W4297206661","doi":"https://doi.org/10.48550/arxiv.2204.06260"},"language":"en","primary_location":{"id":"pmh:oai:arXiv.org:2204.06260","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2204.06260","pdf_url":"https://arxiv.org/pdf/2204.06260","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"volume":null,"type":"preprint","indexed_in":["arxiv","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2204.06260","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5100418547","display_name":"Chen Chen","orcid":"https://orcid.org/0000-0003-1957-6432","observed_orcids":["https://orcid.org/0000-0003-1957-6432"]},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Chen, Chen","raw_affiliation_strings":["School of Computer Science and Engineering , Nanyang Technological University , Singapore"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering , Nanyang Technological University , Singapore","institution_ids":["https://openalex.org/I172675005"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5074544822","display_name":"Yu\u2010Chen Hu","orcid":"https://orcid.org/0000-0002-5055-3645","observed_orcids":["https://orcid.org/0000-0002-5055-3645"]},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Hu, Yuchen","raw_affiliation_strings":["School of Computer Science and Engineering , Nanyang Technological University , Singapore"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering , Nanyang Technological University , Singapore","institution_ids":["https://openalex.org/I172675005"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5001511287","display_name":"Nana Hou","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Hou, Nana","raw_affiliation_strings":["School of Computer Science and Engineering , Nanyang Technological University , Singapore"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering , Nanyang Technological University , Singapore","institution_ids":["https://openalex.org/I172675005"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5110451381","display_name":"Xiaofeng Qi","orcid":null,"observed_orcids":[]},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Qi, Xiaofeng","raw_affiliation_strings":["School of Computer Science and Engineering , Nanyang Technological University , Singapore"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering , Nanyang Technological University , Singapore","institution_ids":["https://openalex.org/I172675005"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5023152132","display_name":"Heqing Zou","orcid":"https://orcid.org/0000-0003-0038-2822","observed_orcids":["https://orcid.org/0000-0003-0038-2822"]},"institutions":[{"id":"https://openalex.org/I172675005","display_name":"Nanyang Technological University","ror":"https://ror.org/02e7b5302","country_code":"SG","type":"education","lineage":["https://openalex.org/I172675005"]}],"countries":["SG"],"is_corresponding":false,"raw_author_name":"Zou, Heqing","raw_affiliation_strings":["School of Computer Science and Engineering , Nanyang Technological University , Singapore"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"School of Computer Science and Engineering , Nanyang Technological University , Singapore","institution_ids":["https://openalex.org/I172675005"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5070872826","display_name":"Eng Siong Chng","orcid":"https://orcid.org/0000-0001-6257-7399","observed_orcids":["https://orcid.org/0000-0001-6257-7399"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Chng, Eng Siong","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"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":9.999999747378752e-05,"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/speech-recognition","display_name":"speech recognition","score":0.9580000042915344},{"id":"https://openalex.org/keywords/word-error-rate","display_name":"word error rate","score":0.8029999732971191},{"id":"https://openalex.org/keywords/teacher-forcing","display_name":"teacher forcing","score":0.5870000123977661},{"id":"https://openalex.org/keywords/reinforcement-learning","display_name":"reinforcement learning","score":0.7049999833106995},{"id":"https://openalex.org/keywords/reward-function-design","display_name":"reward function design","score":0.8629999756813049},{"id":"https://openalex.org/keywords/sequence-to-sequence-models","display_name":"sequence-to-sequence models","score":0.8930000066757202},{"id":"https://openalex.org/keywords/noisy-speech","display_name":"noisy speech","score":0.6650000214576721},{"id":"https://openalex.org/keywords/training-test-mismatch","display_name":"training-test mismatch","score":0.5550000071525574}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7177659273147583},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.6320462226867676},{"id":"https://openalex.org/C2780451532","wikidata":"https://www.wikidata.org/wiki/Q759676","display_name":"Task (project management)","level":2,"score":0.5707250833511353},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.5462668538093567},{"id":"https://openalex.org/C197115733","wikidata":"https://www.wikidata.org/wiki/Q1003136","display_name":"Forcing (mathematics)","level":2,"score":0.541869580745697},{"id":"https://openalex.org/C40969351","wikidata":"https://www.wikidata.org/wiki/Q3516228","display_name":"Word error rate","level":2,"score":0.5382959246635437},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5108144283294678},{"id":"https://openalex.org/C2778112365","wikidata":"https://www.wikidata.org/wiki/Q3511065","display_name":"Sequence (biology)","level":2,"score":0.5096209049224854},{"id":"https://openalex.org/C35639132","wikidata":"https://www.wikidata.org/wiki/Q7452468","display_name":"Sequence labeling","level":3,"score":0.48150262236595154},{"id":"https://openalex.org/C51632099","wikidata":"https://www.wikidata.org/wiki/Q3985153","display_name":"Training set","level":2,"score":0.4168281853199005},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.34816646575927734},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.13406452536582947},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.10162734985351562},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.0},{"id":"https://openalex.org/C201995342","wikidata":"https://www.wikidata.org/wiki/Q682496","display_name":"Systems engineering","level":1,"score":0.0},{"id":"https://openalex.org/C54355233","wikidata":"https://www.wikidata.org/wiki/Q7162","display_name":"Genetics","level":1,"score":0.0},{"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":[{"id":"pmh:oai:arXiv.org:2204.06260","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2204.06260","pdf_url":"https://arxiv.org/pdf/2204.06260","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},{"id":"doi:10.48550/arxiv.2204.06260","doi":"https://doi.org/10.48550/arxiv.2204.06260","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2204.06260","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"Preprint"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2204.06260","doi":null,"is_oa":true,"landing_page_url":"http://arxiv.org/abs/2204.06260","pdf_url":"https://arxiv.org/pdf/2204.06260","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"},"sustainable_development_goals":[],"study_designs":[],"awards":[{"id":"https://openalex.org/G3191157935","display_name":null,"funder_award_id":"AISG-100E-2018-006","funder_id":"https://openalex.org/F4320320709","funder_display_name":"National Research Foundation Singapore"}],"funders":[{"id":"https://openalex.org/F4320320671","display_name":"National Research Foundation","ror":"https://ror.org/05s0g1g46"},{"id":"https://openalex.org/F4320320709","display_name":"National Research Foundation Singapore","ror":"https://ror.org/03cpyc314"}],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W4297206661.pdf","grobid_xml":"https://content.openalex.org/works/W4297206661.grobid-xml"},"referenced_works_count":0,"referenced_works":[],"related_works":["https://openalex.org/W2963144852","https://openalex.org/W2773723600","https://openalex.org/W2951444698","https://openalex.org/W3025712427","https://openalex.org/W4392883629","https://openalex.org/W3095288132","https://openalex.org/W4381327923","https://openalex.org/W3089863108","https://openalex.org/W4409348114","https://openalex.org/W4319862631"],"abstract_inverted_index":{"Although":[0],"automatic":[1],"speech":[2,151],"recognition":[3],"(ASR)":[4],"task":[5],"has":[6,72],"gained":[7],"remarkable":[8],"success":[9],"by":[10,48],"sequence-to-sequence":[11],"models,":[12],"there":[13],"are":[14],"two":[15],"main":[16],"mismatches":[17],"between":[18],"its":[19,77,140],"training":[20,41,93,98,122],"and":[21,124,133,149,153,164],"testing":[22,104],"that":[23,70,157],"might":[24],"lead":[25],"to":[26,36,60,76,95,102,119,139],"performance":[27,45],"degradation:":[28],"1)":[29],"The":[30,56],"typically":[31],"used":[32],"cross-entropy":[33],"criterion":[34,123],"aims":[35],"maximize":[37],"log-likelihood":[38],"of":[39,173],"the":[40,44,61,97,103,121,129,135,154,158,169],"data,":[42],"while":[43],"is":[46],"evaluated":[47],"word":[49],"error":[50],"rate":[51],"(WER),":[52],"not":[53],"log-likelihood;":[54],"2)":[55],"teacher-forcing":[57,132],"method":[58,89],"leads":[59],"dependence":[62],"on":[63,131,146],"ground":[64],"truth":[65],"during":[66],"training,":[67],"which":[68],"means":[69],"model":[71,136],"never":[73],"been":[74],"exposed":[75],"own":[78],"prediction":[79],"before":[80],"testing.":[81],"In":[82],"this":[83],"paper,":[84],"we":[85],"propose":[86],"an":[87],"optimization":[88],"called":[90],"self-critical":[91],"sequence":[92],"(SCST)":[94],"make":[96],"procedure":[99],"much":[100],"closer":[101],"phase.":[105],"As":[106],"a":[107,115],"reinforcement":[108],"learning":[109],"(RL)":[110],"based":[111],"method,":[112],"SCST":[113,160],"utilizes":[114],"customized":[116],"reward":[117],"function":[118],"associate":[120],"WER.":[125,174],"Furthermore,":[126],"it":[127],"removes":[128],"reliance":[130],"harmonizes":[134],"with":[137],"respect":[138],"inference":[141],"procedure.":[142],"We":[143],"conducted":[144],"experiments":[145],"both":[147],"clean":[148],"noisy":[150],"datasets,":[152],"results":[155],"show":[156],"proposed":[159],"respectively":[161],"achieves":[162],"8.7%":[163],"7.8%":[165],"relative":[166],"improvements":[167],"over":[168],"baseline":[170],"in":[171],"terms":[172]},"counts_by_year":[],"updated_date":"2026-09-30T09:27:27.532063","created_date":"2022-09-27T00:00:00"}