{"id":"https://openalex.org/W3197901717","doi":"https://doi.org/10.18653/v1/2021.emnlp-main.832","title":"Beyond Preserved Accuracy: Evaluating Loyalty and Robustness of BERT Compression","display_name":"Beyond Preserved Accuracy: Evaluating Loyalty and Robustness of BERT Compression","publication_year":2021,"publication_date":"2021-01-01","ids":{"openalex":"https://openalex.org/W3197901717","doi":"https://doi.org/10.18653/v1/2021.emnlp-main.832","mag":"3197901717"},"language":"en","primary_location":{"id":"doi:10.18653/v1/2021.emnlp-main.832","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2021.emnlp-main.832","pdf_url":"https://aclanthology.org/2021.emnlp-main.832.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"hybrid","oa_url":"https://aclanthology.org/2021.emnlp-main.832.pdf","any_repository_has_fulltext":false},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5102941860","display_name":"Canwen Xu","orcid":"https://orcid.org/0000-0002-1552-999X"},"institutions":[{"id":"https://openalex.org/I36258959","display_name":"University of California, San Diego","ror":"https://ror.org/0168r3w48","country_code":"US","type":"education","lineage":["https://openalex.org/I36258959"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Canwen Xu","raw_affiliation_strings":["University of California, San Diego"],"affiliations":[{"raw_affiliation_string":"University of California, San Diego","institution_ids":["https://openalex.org/I36258959"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5074508950","display_name":"Wangchunshu Zhou","orcid":"https://orcid.org/0000-0003-4668-3348"},"institutions":[{"id":"https://openalex.org/I97018004","display_name":"Stanford University","ror":"https://ror.org/00f54p054","country_code":"US","type":"education","lineage":["https://openalex.org/I97018004"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Wangchunshu Zhou","raw_affiliation_strings":["Stanford University"],"affiliations":[{"raw_affiliation_string":"Stanford University","institution_ids":["https://openalex.org/I97018004"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5104111369","display_name":"Tao Ge","orcid":null},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Tao Ge","raw_affiliation_strings":["Microsoft Research Asia"],"affiliations":[{"raw_affiliation_string":"Microsoft Research Asia","institution_ids":["https://openalex.org/I4210113369"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100665814","display_name":"Ke Xu","orcid":"https://orcid.org/0000-0003-2587-8517"},"institutions":[{"id":"https://openalex.org/I82880672","display_name":"Beihang University","ror":"https://ror.org/00wk2mp56","country_code":"CN","type":"education","lineage":["https://openalex.org/I82880672"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Ke Xu","raw_affiliation_strings":["Beihang University"],"affiliations":[{"raw_affiliation_string":"Beihang University","institution_ids":["https://openalex.org/I82880672"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5021827617","display_name":"Julian McAuley","orcid":"https://orcid.org/0000-0003-0955-7588"},"institutions":[{"id":"https://openalex.org/I36258959","display_name":"University of California, San Diego","ror":"https://ror.org/0168r3w48","country_code":"US","type":"education","lineage":["https://openalex.org/I36258959"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Julian McAuley","raw_affiliation_strings":["University of California, San Diego"],"affiliations":[{"raw_affiliation_string":"University of California, San Diego","institution_ids":["https://openalex.org/I36258959"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5014662947","display_name":"Furu Wei","orcid":"https://orcid.org/0000-0002-7810-5852"},"institutions":[{"id":"https://openalex.org/I4210113369","display_name":"Microsoft Research Asia (China)","ror":"https://ror.org/0300m5276","country_code":"CN","type":"company","lineage":["https://openalex.org/I1290206253","https://openalex.org/I4210113369"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Furu Wei","raw_affiliation_strings":["Microsoft Research Asia"],"affiliations":[{"raw_affiliation_string":"Microsoft Research Asia","institution_ids":["https://openalex.org/I4210113369"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":6,"corresponding_author_ids":["https://openalex.org/A5102941860"],"corresponding_institution_ids":["https://openalex.org/I36258959"],"apc_list":null,"apc_paid":null,"fwci":3.1715,"has_fulltext":true,"cited_by_count":26,"citation_normalized_percentile":{"value":0.93427621,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":89,"max":100},"biblio":{"volume":null,"issue":null,"first_page":"10653","last_page":"10659"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.9997000098228455,"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/T11689","display_name":"Adversarial Robustness in Machine Learning","score":0.9997000098228455,"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":0.9990000128746033,"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"}},{"id":"https://openalex.org/T12026","display_name":"Explainable Artificial Intelligence (XAI)","score":0.9912999868392944,"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/computer-science","display_name":"Computer science","score":0.7611638307571411},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.7495028376579285},{"id":"https://openalex.org/keywords/loyalty","display_name":"Loyalty","score":0.6589483618736267},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.49612316489219666},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.4803946614265442},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.4327138364315033},{"id":"https://openalex.org/keywords/speech-recognition","display_name":"Speech recognition","score":0.334930956363678},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.2612948417663574}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7611638307571411},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.7495028376579285},{"id":"https://openalex.org/C2776967331","wikidata":"https://www.wikidata.org/wiki/Q1132131","display_name":"Loyalty","level":2,"score":0.6589483618736267},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.49612316489219666},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.4803946614265442},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.4327138364315033},{"id":"https://openalex.org/C28490314","wikidata":"https://www.wikidata.org/wiki/Q189436","display_name":"Speech recognition","level":1,"score":0.334930956363678},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2612948417663574},{"id":"https://openalex.org/C17744445","wikidata":"https://www.wikidata.org/wiki/Q36442","display_name":"Political science","level":0,"score":0.0},{"id":"https://openalex.org/C104317684","wikidata":"https://www.wikidata.org/wiki/Q7187","display_name":"Gene","level":2,"score":0.0},{"id":"https://openalex.org/C55493867","wikidata":"https://www.wikidata.org/wiki/Q7094","display_name":"Biochemistry","level":1,"score":0.0},{"id":"https://openalex.org/C199539241","wikidata":"https://www.wikidata.org/wiki/Q7748","display_name":"Law","level":1,"score":0.0},{"id":"https://openalex.org/C185592680","wikidata":"https://www.wikidata.org/wiki/Q2329","display_name":"Chemistry","level":0,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.18653/v1/2021.emnlp-main.832","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2021.emnlp-main.832","pdf_url":"https://aclanthology.org/2021.emnlp-main.832.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.18653/v1/2021.emnlp-main.832","is_oa":true,"landing_page_url":"https://doi.org/10.18653/v1/2021.emnlp-main.832","pdf_url":"https://aclanthology.org/2021.emnlp-main.832.pdf","source":{"id":"https://openalex.org/S4363608991","display_name":"Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing","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":"cc-by","license_id":"https://openalex.org/licenses/cc-by","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W3197901717.pdf","grobid_xml":"https://content.openalex.org/works/W3197901717.grobid-xml"},"referenced_works_count":54,"referenced_works":["https://openalex.org/W1821462560","https://openalex.org/W2251939518","https://openalex.org/W2626967530","https://openalex.org/W2766839578","https://openalex.org/W2887603965","https://openalex.org/W2896457183","https://openalex.org/W2923014074","https://openalex.org/W2963310665","https://openalex.org/W2963341956","https://openalex.org/W2963542245","https://openalex.org/W2963809228","https://openalex.org/W2963846996","https://openalex.org/W2964212410","https://openalex.org/W2965373594","https://openalex.org/W2965595599","https://openalex.org/W2969601108","https://openalex.org/W2970454332","https://openalex.org/W2970565456","https://openalex.org/W2970597249","https://openalex.org/W2970699417","https://openalex.org/W2973727699","https://openalex.org/W2975429091","https://openalex.org/W2978017171","https://openalex.org/W2979826702","https://openalex.org/W2997279057","https://openalex.org/W2998183051","https://openalex.org/W3000076995","https://openalex.org/W3008374555","https://openalex.org/W3012223454","https://openalex.org/W3014936620","https://openalex.org/W3034292689","https://openalex.org/W3034457371","https://openalex.org/W3035030897","https://openalex.org/W3035038672","https://openalex.org/W3082274269","https://openalex.org/W3091818438","https://openalex.org/W3098824823","https://openalex.org/W3101163004","https://openalex.org/W3101248447","https://openalex.org/W3104128335","https://openalex.org/W3104738015","https://openalex.org/W3105966348","https://openalex.org/W3115348206","https://openalex.org/W3123662205","https://openalex.org/W3133702157","https://openalex.org/W3167791066","https://openalex.org/W3177265267","https://openalex.org/W3196295870","https://openalex.org/W3214897310","https://openalex.org/W4287777801","https://openalex.org/W4288089799","https://openalex.org/W4288256350","https://openalex.org/W4288419306","https://openalex.org/W4297816198"],"related_works":["https://openalex.org/W2961085424","https://openalex.org/W4306674287","https://openalex.org/W4224009465","https://openalex.org/W4286629047","https://openalex.org/W4306321456","https://openalex.org/W4285260836","https://openalex.org/W3046775127","https://openalex.org/W3170094116","https://openalex.org/W4205958290","https://openalex.org/W3209574120"],"abstract_inverted_index":{"Recent":[0],"studies":[1],"on":[2],"compression":[3,53,78],"of":[4,52],"pretrained":[5],"language":[6],"models":[7],"(e.g.,":[8],"BERT)":[9],"usually":[10],"use":[11],"preserved":[12],"accuracy":[13],"as":[14],"the":[15,42,50],"metric":[16],"for":[17],"evaluation.":[18],"In":[19],"this":[20],"paper,":[21],"we":[22,80],"propose":[23],"two":[24],"new":[25],"metrics,":[26],"label":[27],"loyalty":[28,31,72,89],"and":[29,67,73,90],"probability":[30],"that":[32],"measure":[33],"how":[34],"closely":[35],"a":[36,82],"compressed":[37],"model":[38,44],"(i.e.,":[39,45],"student)":[40],"mimics":[41],"original":[43],"teacher).":[46],"We":[47,61],"also":[48],"explore":[49],"effect":[51],"with":[54,71],"regard":[55],"to":[56,85],"robustness":[57],"under":[58],"adversarial":[59],"attacks.":[60],"benchmark":[62],"quantization,":[63],"pruning,":[64],"knowledge":[65],"distillation":[66],"progressive":[68],"module":[69],"replacing":[70],"robustness.":[74,91],"By":[75],"combining":[76],"multiple":[77],"techniques,":[79],"provide":[81],"practical":[83],"strategy":[84],"achieve":[86],"better":[87],"accuracy,":[88]},"counts_by_year":[{"year":2026,"cited_by_count":1},{"year":2024,"cited_by_count":2},{"year":2023,"cited_by_count":15},{"year":2022,"cited_by_count":7},{"year":2021,"cited_by_count":1}],"updated_date":"2026-04-06T07:47:59.780226","created_date":"2025-10-10T00:00:00"}