{"id":"https://openalex.org/W4285295266","doi":"https://doi.org/10.1109/taslp.2022.3180678","title":"Attention Analysis and Calibration for Transformer in Natural Language Generation","display_name":"Attention Analysis and Calibration for Transformer in Natural Language Generation","publication_year":2022,"publication_date":"2022-01-01","ids":{"openalex":"https://openalex.org/W4285295266","doi":"https://doi.org/10.1109/taslp.2022.3180678"},"language":"en","primary_location":{"id":"doi:10.1109/taslp.2022.3180678","is_oa":false,"landing_page_url":"https://doi.org/10.1109/taslp.2022.3180678","pdf_url":null,"source":{"id":"https://openalex.org/S4210169297","display_name":"IEEE/ACM Transactions on Audio Speech and Language Processing","issn_l":"2329-9290","issn":["2329-9290","2329-9304"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE/ACM Transactions on Audio, Speech, and Language Processing","raw_type":"journal-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/A5009484350","display_name":"Yu Lu","orcid":"https://orcid.org/0000-0001-9108-2619"},"institutions":[{"id":"https://openalex.org/I19820366","display_name":"Chinese Academy of Sciences","ror":"https://ror.org/034t30j35","country_code":"CN","type":"government","lineage":["https://openalex.org/I19820366"]},{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]},{"id":"https://openalex.org/I4210112150","display_name":"Institute of Automation","ror":"https://ror.org/022c3hy66","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210112150"]},{"id":"https://openalex.org/I4210165038","display_name":"University of Chinese Academy of Sciences","ror":"https://ror.org/05qbk4x57","country_code":"CN","type":"education","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210165038"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yu Lu","raw_affiliation_strings":["National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences (CAS), Beijing, China","School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0001-9108-2619","affiliations":[{"raw_affiliation_string":"National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences (CAS), Beijing, China","institution_ids":["https://openalex.org/I19820366","https://openalex.org/I4210112150"]},{"raw_affiliation_string":"School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China","institution_ids":["https://openalex.org/I4210100255","https://openalex.org/I4210165038"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100319572","display_name":"Jiajun Zhang","orcid":"https://orcid.org/0000-0001-5293-7434"},"institutions":[{"id":"https://openalex.org/I19820366","display_name":"Chinese Academy of Sciences","ror":"https://ror.org/034t30j35","country_code":"CN","type":"government","lineage":["https://openalex.org/I19820366"]},{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]},{"id":"https://openalex.org/I4210112150","display_name":"Institute of Automation","ror":"https://ror.org/022c3hy66","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210112150"]},{"id":"https://openalex.org/I4210165038","display_name":"University of Chinese Academy of Sciences","ror":"https://ror.org/05qbk4x57","country_code":"CN","type":"education","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210165038"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jiajun Zhang","raw_affiliation_strings":["National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences (CAS), Beijing, China","School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0001-5293-7434","affiliations":[{"raw_affiliation_string":"National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences (CAS), Beijing, China","institution_ids":["https://openalex.org/I19820366","https://openalex.org/I4210112150"]},{"raw_affiliation_string":"School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China","institution_ids":["https://openalex.org/I4210100255","https://openalex.org/I4210165038"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101872305","display_name":"Jiali Zeng","orcid":"https://orcid.org/0000-0003-0808-9890"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Jiali Zeng","raw_affiliation_strings":["Department Tencent Cloud Xiaowei, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0003-0808-9890","affiliations":[{"raw_affiliation_string":"Department Tencent Cloud Xiaowei, Beijing, China","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5091839641","display_name":"Shuangzhi Wu","orcid":"https://orcid.org/0000-0002-6371-9720"},"institutions":[{"id":"https://openalex.org/I2250653659","display_name":"Tencent (China)","ror":"https://ror.org/00hhjss72","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250653659"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shuangzhi Wu","raw_affiliation_strings":["Department Tencent Cloud Xiaowei, Beijing, China"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Department Tencent Cloud Xiaowei, Beijing, China","institution_ids":["https://openalex.org/I2250653659"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5015785439","display_name":"Chengqing Zong","orcid":"https://orcid.org/0000-0002-9864-3818"},"institutions":[{"id":"https://openalex.org/I19820366","display_name":"Chinese Academy of Sciences","ror":"https://ror.org/034t30j35","country_code":"CN","type":"government","lineage":["https://openalex.org/I19820366"]},{"id":"https://openalex.org/I4210100255","display_name":"Beijing Academy of Artificial Intelligence","ror":"https://ror.org/016a74861","country_code":"CN","type":"other","lineage":["https://openalex.org/I4210100255"]},{"id":"https://openalex.org/I4210112150","display_name":"Institute of Automation","ror":"https://ror.org/022c3hy66","country_code":"CN","type":"facility","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210112150"]},{"id":"https://openalex.org/I4210165038","display_name":"University of Chinese Academy of Sciences","ror":"https://ror.org/05qbk4x57","country_code":"CN","type":"education","lineage":["https://openalex.org/I19820366","https://openalex.org/I4210165038"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Chengqing Zong","raw_affiliation_strings":["National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences (CAS), Beijing, China","School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China"],"raw_orcid":"https://orcid.org/0000-0002-9864-3818","affiliations":[{"raw_affiliation_string":"National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences (CAS), Beijing, China","institution_ids":["https://openalex.org/I19820366","https://openalex.org/I4210112150"]},{"raw_affiliation_string":"School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing, China","institution_ids":["https://openalex.org/I4210100255","https://openalex.org/I4210165038"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":5,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.8122,"has_fulltext":false,"cited_by_count":7,"citation_normalized_percentile":{"value":0.76002879,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":94,"max":97},"biblio":{"volume":"30","issue":null,"first_page":"1927","last_page":"1938"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10028","display_name":"Topic Modeling","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/T10028","display_name":"Topic Modeling","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/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/T11714","display_name":"Multimodal Machine Learning Applications","score":0.9927999973297119,"subfield":{"id":"https://openalex.org/subfields/1707","display_name":"Computer Vision and Pattern Recognition"},"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/computer-science","display_name":"Computer science","score":0.7970762252807617},{"id":"https://openalex.org/keywords/machine-translation","display_name":"Machine translation","score":0.7637301087379456},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.6571102738380432},{"id":"https://openalex.org/keywords/automatic-summarization","display_name":"Automatic summarization","score":0.6493160724639893},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.6099115610122681},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"Natural language processing","score":0.5838569402694702},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.5672727823257446},{"id":"https://openalex.org/keywords/phrase","display_name":"Phrase","score":0.47508302330970764},{"id":"https://openalex.org/keywords/natural-language-generation","display_name":"Natural language generation","score":0.4369350075721741},{"id":"https://openalex.org/keywords/natural-language","display_name":"Natural language","score":0.3248109221458435}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7970762252807617},{"id":"https://openalex.org/C203005215","wikidata":"https://www.wikidata.org/wiki/Q79798","display_name":"Machine translation","level":2,"score":0.7637301087379456},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6571102738380432},{"id":"https://openalex.org/C170858558","wikidata":"https://www.wikidata.org/wiki/Q1394144","display_name":"Automatic summarization","level":2,"score":0.6493160724639893},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.6099115610122681},{"id":"https://openalex.org/C204321447","wikidata":"https://www.wikidata.org/wiki/Q30642","display_name":"Natural language processing","level":1,"score":0.5838569402694702},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5672727823257446},{"id":"https://openalex.org/C2776224158","wikidata":"https://www.wikidata.org/wiki/Q187931","display_name":"Phrase","level":2,"score":0.47508302330970764},{"id":"https://openalex.org/C2776187449","wikidata":"https://www.wikidata.org/wiki/Q1513879","display_name":"Natural language generation","level":3,"score":0.4369350075721741},{"id":"https://openalex.org/C195324797","wikidata":"https://www.wikidata.org/wiki/Q33742","display_name":"Natural language","level":2,"score":0.3248109221458435},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.0},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/taslp.2022.3180678","is_oa":false,"landing_page_url":"https://doi.org/10.1109/taslp.2022.3180678","pdf_url":null,"source":{"id":"https://openalex.org/S4210169297","display_name":"IEEE/ACM Transactions on Audio Speech and Language Processing","issn_l":"2329-9290","issn":["2329-9290","2329-9304"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319808","host_organization_name":"Institute of Electrical and Electronics Engineers","host_organization_lineage":["https://openalex.org/P4310319808"],"host_organization_lineage_names":["Institute of Electrical and Electronics Engineers"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE/ACM Transactions on Audio, Speech, and Language Processing","raw_type":"journal-article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Peace, Justice and strong institutions","id":"https://metadata.un.org/sdg/16","score":0.6100000143051147}],"awards":[{"id":"https://openalex.org/G4874090527","display_name":null,"funder_award_id":"62122088","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G5704739355","display_name":null,"funder_award_id":"62006224","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"},{"id":"https://openalex.org/G7517328767","display_name":null,"funder_award_id":"U1836221","funder_id":"https://openalex.org/F4320321001","funder_display_name":"National Natural Science Foundation of China"}],"funders":[{"id":"https://openalex.org/F4320321001","display_name":"National Natural Science Foundation of China","ror":"https://ror.org/01h0zpd94"}],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":58,"referenced_works":["https://openalex.org/W1902237438","https://openalex.org/W2101105183","https://openalex.org/W2105482032","https://openalex.org/W2124807415","https://openalex.org/W2141532438","https://openalex.org/W2146502635","https://openalex.org/W2467173223","https://openalex.org/W2525778437","https://openalex.org/W2600144439","https://openalex.org/W2606462007","https://openalex.org/W2606974598","https://openalex.org/W2808413133","https://openalex.org/W2891635683","https://openalex.org/W2903268415","https://openalex.org/W2906152891","https://openalex.org/W2912351236","https://openalex.org/W2916755661","https://openalex.org/W2946794439","https://openalex.org/W2949161734","https://openalex.org/W2950768109","https://openalex.org/W2952682849","https://openalex.org/W2962708992","https://openalex.org/W2962784628","https://openalex.org/W2962965405","https://openalex.org/W2963260202","https://openalex.org/W2963333747","https://openalex.org/W2963506925","https://openalex.org/W2963598809","https://openalex.org/W2963929190","https://openalex.org/W2964165364","https://openalex.org/W2970045405","https://openalex.org/W2970726176","https://openalex.org/W2972498556","https://openalex.org/W2972918955","https://openalex.org/W2996061341","https://openalex.org/W3031696893","https://openalex.org/W3035281110","https://openalex.org/W3035691816","https://openalex.org/W3037212200","https://openalex.org/W3101609372","https://openalex.org/W3104881680","https://openalex.org/W3106185885","https://openalex.org/W3173506780","https://openalex.org/W3174556939","https://openalex.org/W6681435938","https://openalex.org/W6682026795","https://openalex.org/W6682631176","https://openalex.org/W6688536550","https://openalex.org/W6719786876","https://openalex.org/W6725207838","https://openalex.org/W6726804950","https://openalex.org/W6727690538","https://openalex.org/W6737778391","https://openalex.org/W6739901393","https://openalex.org/W6745761326","https://openalex.org/W6764072591","https://openalex.org/W6767868144","https://openalex.org/W6783672905"],"related_works":["https://openalex.org/W4284703357","https://openalex.org/W2949968076","https://openalex.org/W2935811960","https://openalex.org/W2969407538","https://openalex.org/W2955859849","https://openalex.org/W2152921782","https://openalex.org/W382594479","https://openalex.org/W2470045054","https://openalex.org/W2575772232","https://openalex.org/W2151245229"],"abstract_inverted_index":{"Attention":[0],"mechanism":[1],"has":[2],"been":[3],"ubiquitous":[4],"in":[5,48,84],"neural":[6,67],"machine":[7,68],"translation":[8,114,194],"by":[9,51,175],"dynamically":[10],"selecting":[11],"relevant":[12],"contexts":[13],"for":[14],"different":[15],"translations.":[16],"Apart":[17],"from":[18,157],"performance":[19,188],"gains,":[20],"attention":[21,54,75,118,134,152,162,174,178],"weights":[22,55,153],"assigned":[23],"to":[24,30,37,65,76,100,111,169,179,185],"input":[25],"tokens":[26,34],"are":[27],"often":[28],"utilized":[29],"explain":[31],"that":[32,122,201],"high-attention":[33],"contribute":[35],"more":[36,177],"the":[38,107,112,116,131,136,140,149,158,171],"prediction.":[39],"However,":[40],"many":[41],"works":[42],"question":[43,64],"whether":[44,106],"this":[45,63,85],"assumption":[46],"holds":[47],"text":[49,197],"classification":[50],"manually":[52],"manipulating":[53],"and":[56,143,154,196],"observing":[57],"decision":[58],"flips.":[59],"This":[60],"article":[61],"extends":[62],"Transformer-based":[66],"translation,":[69],"which":[70,94,127],"heavily":[71],"relies":[72],"on":[73,130,192],"cross-lingual":[74,172],"produce":[77],"accurate":[78],"translations":[79],"but":[80],"is":[81],"relatively":[82],"understudied":[83],"context.":[86],"We":[87,103,120,146],"first":[88],"design":[89],"a":[90,186],"mask":[91],"perturbation":[92],"model":[93,101],"automatically":[95],"assesses":[96],"each":[97],"input’s":[98],"contribution":[99],"outputs.":[102],"then":[104],"test":[105],"token":[108],"contributing":[109],"most":[110],"current":[113,141],"receives":[115],"highest":[117],"weight.":[119],"find":[121],"it":[123],"sometimes":[124],"does":[125],"not,":[126],"closely":[128],"depends":[129],"entropy":[132],"of":[133,139,160],"weights,":[135],"syntactic":[137],"role":[138],"generation,":[142],"language":[144],"pairs.":[145],"also":[147],"rethink":[148],"discrepancy":[150],"between":[151],"word":[155],"alignments":[156],"view":[159],"unreliable":[161],"weights.":[163],"Our":[164],"observations":[165],"further":[166],"motivate":[167],"us":[168],"calibrate":[170],"multi-head":[173],"attaching":[176],"indispensable":[180],"tokens,":[181],"whose":[182],"removal":[183],"leads":[184],"dramatic":[187],"drop.":[189],"Empirical":[190],"experiments":[191],"different-scale":[193],"tasks":[195,199],"summarization":[198],"demonstrate":[200],"our":[202],"calibration":[203],"methods":[204],"significantly":[205],"outperform":[206],"strong":[207],"baselines.":[208]},"counts_by_year":[{"year":2025,"cited_by_count":3},{"year":2024,"cited_by_count":2},{"year":2023,"cited_by_count":2}],"updated_date":"2026-07-22T07:51:19.307946","created_date":"2025-10-10T00:00:00"}