{"id":"https://openalex.org/W3120174600","doi":"https://doi.org/10.1109/ijcnn52387.2021.9533563","title":"Of Non-Linearity and Commutativity in BERT","display_name":"Of Non-Linearity and Commutativity in BERT","publication_year":2021,"publication_date":"2021-07-18","ids":{"openalex":"https://openalex.org/W3120174600","doi":"https://doi.org/10.1109/ijcnn52387.2021.9533563","mag":"3120174600"},"language":"en","primary_location":{"id":"doi:10.1109/ijcnn52387.2021.9533563","is_oa":false,"landing_page_url":"https://doi.org/10.1109/ijcnn52387.2021.9533563","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2021 International Joint Conference on Neural Networks (IJCNN)","raw_type":"proceedings-article"},"type":"conference-paper","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2101.04547","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101135146","display_name":"Sumu Zhao","orcid":null},"institutions":[{"id":"https://openalex.org/I35440088","display_name":"ETH Zurich","ror":"https://ror.org/05a28rw58","country_code":"CH","type":"education","lineage":["https://openalex.org/I2799323385","https://openalex.org/I35440088"]}],"countries":["CH"],"is_corresponding":false,"raw_author_name":"Sumu Zhao","raw_affiliation_strings":["ETH Zurich, Switzerland, Switzerland"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ETH Zurich, Switzerland, Switzerland","institution_ids":["https://openalex.org/I35440088"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5089481900","display_name":"Dami\u00e1n Pascual","orcid":"https://orcid.org/0000-0002-7018-5368"},"institutions":[{"id":"https://openalex.org/I35440088","display_name":"ETH Zurich","ror":"https://ror.org/05a28rw58","country_code":"CH","type":"education","lineage":["https://openalex.org/I2799323385","https://openalex.org/I35440088"]}],"countries":["CH"],"is_corresponding":false,"raw_author_name":"Damian Pascual","raw_affiliation_strings":["ETH Zurich, Switzerland, Switzerland","ETH Z\u00fcrich"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ETH Zurich, Switzerland, Switzerland","institution_ids":["https://openalex.org/I35440088"]},{"raw_affiliation_string":"ETH Z\u00fcrich","institution_ids":["https://openalex.org/I35440088"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5047790045","display_name":"Gino Brunner","orcid":"https://orcid.org/0000-0002-4341-2940"},"institutions":[{"id":"https://openalex.org/I35440088","display_name":"ETH Zurich","ror":"https://ror.org/05a28rw58","country_code":"CH","type":"education","lineage":["https://openalex.org/I2799323385","https://openalex.org/I35440088"]}],"countries":["CH"],"is_corresponding":false,"raw_author_name":"Gino Brunner","raw_affiliation_strings":["ETH Zurich, Switzerland, Switzerland","ETH Z\u00fcrich"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ETH Zurich, Switzerland, Switzerland","institution_ids":["https://openalex.org/I35440088"]},{"raw_affiliation_string":"ETH Z\u00fcrich","institution_ids":["https://openalex.org/I35440088"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5078339613","display_name":"Roger Wattenhofer","orcid":"https://orcid.org/0000-0002-6339-3134"},"institutions":[{"id":"https://openalex.org/I35440088","display_name":"ETH Zurich","ror":"https://ror.org/05a28rw58","country_code":"CH","type":"education","lineage":["https://openalex.org/I2799323385","https://openalex.org/I35440088"]}],"countries":["CH"],"is_corresponding":false,"raw_author_name":"Roger Wattenhofer","raw_affiliation_strings":["ETH Zurich, Switzerland, Switzerland","ETH Z\u00fcrich"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"ETH Zurich, Switzerland, Switzerland","institution_ids":["https://openalex.org/I35440088"]},{"raw_affiliation_string":"ETH Z\u00fcrich","institution_ids":["https://openalex.org/I35440088"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":1,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I35440088"],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":1,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":"1","last_page":"8"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9972000122070312,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},"topics":[{"id":"https://openalex.org/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9972000122070312,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9843000173568726,"subfield":{"id":"https://openalex.org/subfields/2208","display_name":"Electrical and Electronic Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}},{"id":"https://openalex.org/T11948","display_name":"Machine Learning in Materials Science","score":0.9830999970436096,"subfield":{"id":"https://openalex.org/subfields/2505","display_name":"Materials Chemistry"},"field":{"id":"https://openalex.org/fields/25","display_name":"Materials Science"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.7829976081848145},{"id":"https://openalex.org/keywords/linearity","display_name":"Linearity","score":0.6854668855667114},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.614719033241272},{"id":"https://openalex.org/keywords/commutative-property","display_name":"Commutative property","score":0.5941254496574402},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.5725125074386597},{"id":"https://openalex.org/keywords/fuzzy-logic","display_name":"Fuzzy logic","score":0.46782058477401733},{"id":"https://openalex.org/keywords/electronic-engineering","display_name":"Electronic engineering","score":0.308501660823822},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.21178147196769714},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.20884421467781067},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.2011343240737915},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.13868460059165955},{"id":"https://openalex.org/keywords/discrete-mathematics","display_name":"Discrete mathematics","score":0.09562543034553528}],"concepts":[{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.7829976081848145},{"id":"https://openalex.org/C77170095","wikidata":"https://www.wikidata.org/wiki/Q1753188","display_name":"Linearity","level":2,"score":0.6854668855667114},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.614719033241272},{"id":"https://openalex.org/C183778304","wikidata":"https://www.wikidata.org/wiki/Q165474","display_name":"Commutative property","level":2,"score":0.5941254496574402},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.5725125074386597},{"id":"https://openalex.org/C58166","wikidata":"https://www.wikidata.org/wiki/Q224821","display_name":"Fuzzy logic","level":2,"score":0.46782058477401733},{"id":"https://openalex.org/C24326235","wikidata":"https://www.wikidata.org/wiki/Q126095","display_name":"Electronic engineering","level":1,"score":0.308501660823822},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.21178147196769714},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.20884421467781067},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.2011343240737915},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.13868460059165955},{"id":"https://openalex.org/C118615104","wikidata":"https://www.wikidata.org/wiki/Q121416","display_name":"Discrete mathematics","level":1,"score":0.09562543034553528},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.0},{"id":"https://openalex.org/C142362112","wikidata":"https://www.wikidata.org/wiki/Q735","display_name":"Art","level":0,"score":0.0},{"id":"https://openalex.org/C153349607","wikidata":"https://www.wikidata.org/wiki/Q36649","display_name":"Visual arts","level":1,"score":0.0}],"mesh":[],"locations_count":4,"locations":[{"id":"doi:10.1109/ijcnn52387.2021.9533563","is_oa":false,"landing_page_url":"https://doi.org/10.1109/ijcnn52387.2021.9533563","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2021 International Joint Conference on Neural Networks (IJCNN)","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2101.04547","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2101.04547","pdf_url":"https://arxiv.org/pdf/2101.04547","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":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},{"id":"doi:10.48550/arxiv.2101.04547","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2101.04547","pdf_url":null,"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":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"Preprint"},{"id":"mag:3120174600","is_oa":false,"landing_page_url":null,"pdf_url":null,"source":null,"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":null}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2101.04547","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2101.04547","pdf_url":"https://arxiv.org/pdf/2101.04547","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":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/11","score":0.7599999904632568,"display_name":"Sustainable cities and communities"}],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W3120174600.pdf","grobid_xml":"https://content.openalex.org/works/W3120174600.grobid-xml"},"referenced_works_count":63,"referenced_works":["https://openalex.org/W1566289585","https://openalex.org/W2250539671","https://openalex.org/W2462831000","https://openalex.org/W2866343820","https://openalex.org/W2899663614","https://openalex.org/W2908854766","https://openalex.org/W2934842096","https://openalex.org/W2946417913","https://openalex.org/W2946794439","https://openalex.org/W2948947170","https://openalex.org/W2950813464","https://openalex.org/W2953307569","https://openalex.org/W2963310665","https://openalex.org/W2963341956","https://openalex.org/W2963403868","https://openalex.org/W2965373594","https://openalex.org/W2970120757","https://openalex.org/W2970565456","https://openalex.org/W2970900903","https://openalex.org/W2972324944","https://openalex.org/W2973154008","https://openalex.org/W2973291083","https://openalex.org/W2978017171","https://openalex.org/W2979826702","https://openalex.org/W2981852735","https://openalex.org/W2988217457","https://openalex.org/W2991265431","https://openalex.org/W2995015695","https://openalex.org/W2995446988","https://openalex.org/W2996428491","https://openalex.org/W3000514857","https://openalex.org/W3006683367","https://openalex.org/W3006881356","https://openalex.org/W3015233032","https://openalex.org/W3030163527","https://openalex.org/W3034487470","https://openalex.org/W3035691519","https://openalex.org/W3104350794","https://openalex.org/W3153147196","https://openalex.org/W6692305950","https://openalex.org/W6739901393","https://openalex.org/W6750615492","https://openalex.org/W6753640285","https://openalex.org/W6755207826","https://openalex.org/W6755977528","https://openalex.org/W6757635932","https://openalex.org/W6762945437","https://openalex.org/W6764072591","https://openalex.org/W6766673545","https://openalex.org/W6767563556","https://openalex.org/W6767674921","https://openalex.org/W6768021236","https://openalex.org/W6768061436","https://openalex.org/W6768851824","https://openalex.org/W6769243733","https://openalex.org/W6769627184","https://openalex.org/W6770357686","https://openalex.org/W6771626834","https://openalex.org/W6773996589","https://openalex.org/W6774054309","https://openalex.org/W6775921738","https://openalex.org/W6776129198","https://openalex.org/W6999611261"],"related_works":["https://openalex.org/W3214210379","https://openalex.org/W3012835873","https://openalex.org/W3108320661","https://openalex.org/W1556430621","https://openalex.org/W3037828148","https://openalex.org/W3094002026","https://openalex.org/W135770067","https://openalex.org/W124309221","https://openalex.org/W2150199381","https://openalex.org/W2919038581","https://openalex.org/W2949982775","https://openalex.org/W3181171613","https://openalex.org/W2028702657","https://openalex.org/W2056380766","https://openalex.org/W2182833056","https://openalex.org/W3037787599","https://openalex.org/W2189336758","https://openalex.org/W3206632035","https://openalex.org/W2183408553","https://openalex.org/W2766548934"],"abstract_inverted_index":{"In":[0],"this":[1],"work":[2],"we":[3,19,35,89,127],"provide":[4],"new":[5],"insights":[6],"into":[7],"the":[8,25,40,51,91,101,133,141],"transformer":[9,148],"architecture,":[10],"and":[11,54,73,97,146],"in":[12,86,95,110],"particular,":[13],"its":[14],"best-known":[15],"variant,":[16],"BERT.":[17],"First,":[18],"propose":[20],"a":[21,84,111,138],"method":[22],"to":[23,132],"measure":[24],"degree":[26],"of":[27,29,32,50,144],"non-linearity":[28],"different":[30],"elements":[31],"transformers.":[33],"Next,":[34],"focus":[36],"our":[37],"investigation":[38],"on":[39],"feed-forward":[41],"networks":[42],"(FFN)":[43],"inside":[44],"transformers,":[45],"which":[46,126],"contain":[47],"two":[48],"thirds":[49],"model":[52],"parameters":[53],"have":[55],"so":[56],"far":[57],"not":[58],"received":[59],"much":[60],"attention.":[61],"We":[62],"find":[63,128],"that":[64,74,117],"FFNs":[65],"are":[66],"an":[67,120],"inefficient":[68],"yet":[69],"important":[70],"architectural":[71],"element":[72],"they":[75,107],"cannot":[76],"simply":[77],"be":[78],"replaced":[79],"by":[80],"attention":[81],"blocks":[82],"without":[83],"degradation":[85],"performance.":[87],"Moreover,":[88],"study":[90],"interactions":[92],"between":[93],"layers":[94,102],"BERT":[96,118],"show":[98],"that,":[99],"while":[100],"exhibit":[103],"some":[104],"hierarchical":[105],"structure,":[106],"extract":[108],"features":[109],"fuzzy":[112],"manner.":[113],"Our":[114],"results":[115],"suggest":[116],"has":[119],"inductive":[121],"bias":[122],"towards":[123],"layer":[124],"commutativity,":[125],"is":[129],"mainly":[130],"due":[131],"skip":[134],"connections.":[135],"This":[136],"provides":[137],"justification":[139],"for":[140],"strong":[142],"performance":[143],"recurrent":[145],"weight-shared":[147],"models.":[149]},"counts_by_year":[{"year":2021,"cited_by_count":1}],"updated_date":"2026-07-14T23:27:15.235271","created_date":"2025-10-10T00:00:00"}