{"id":"https://openalex.org/W4312630812","doi":"https://doi.org/10.1109/tpami.2022.3231442","title":"What Makes for Good Tokenizers in Vision Transformer?","display_name":"What Makes for Good Tokenizers in Vision Transformer?","publication_year":2022,"publication_date":"2022-12-22","ids":{"openalex":"https://openalex.org/W4312630812","doi":"https://doi.org/10.1109/tpami.2022.3231442","pmid":"https://pubmed.ncbi.nlm.nih.gov/37015534"},"language":"en","primary_location":{"id":"doi:10.1109/tpami.2022.3231442","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpami.2022.3231442","pdf_url":null,"source":{"id":"https://openalex.org/S199944782","display_name":"IEEE Transactions on Pattern Analysis and Machine Intelligence","issn_l":"0162-8828","issn":["0162-8828","1939-3539","2160-9292"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320439","host_organization_name":"IEEE Computer Society","host_organization_lineage":["https://openalex.org/P4310320439","https://openalex.org/P4310319808"],"host_organization_lineage_names":["IEEE Computer Society","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 Transactions on Pattern Analysis and Machine Intelligence","raw_type":"journal-article"},"type":"article","indexed_in":["crossref","pubmed"],"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/A5034240662","display_name":"Shengju Qian","orcid":"https://orcid.org/0000-0002-3386-0336"},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Shengju Qian","raw_affiliation_strings":["Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0002-3386-0336","affiliations":[{"raw_affiliation_string":"Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5019191351","display_name":"Yi Zhu","orcid":"https://orcid.org/0000-0002-6482-6712"},"institutions":[{"id":"https://openalex.org/I1311688040","display_name":"Amazon (United States)","ror":"https://ror.org/04mv4n011","country_code":"US","type":"company","lineage":["https://openalex.org/I1311688040"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Yi Zhu","raw_affiliation_strings":["Amazon AI, Amazon.com Inc, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Amazon AI, Amazon.com Inc, USA","institution_ids":["https://openalex.org/I1311688040"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100336606","display_name":"Wenbo Li","orcid":"https://orcid.org/0000-0003-4604-778X"},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Wenbo Li","raw_affiliation_strings":["Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0003-4604-778X","affiliations":[{"raw_affiliation_string":"Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5045408107","display_name":"Mu Li","orcid":"https://orcid.org/0000-0003-0193-8386"},"institutions":[{"id":"https://openalex.org/I1311688040","display_name":"Amazon (United States)","ror":"https://ror.org/04mv4n011","country_code":"US","type":"company","lineage":["https://openalex.org/I1311688040"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Mu Li","raw_affiliation_strings":["Amazon AI, Amazon.com Inc, USA"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Amazon AI, Amazon.com Inc, USA","institution_ids":["https://openalex.org/I1311688040"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5052856441","display_name":"Jiaya Jia","orcid":"https://orcid.org/0000-0002-1246-553X"},"institutions":[{"id":"https://openalex.org/I177725633","display_name":"Chinese University of Hong Kong","ror":"https://ror.org/00t33hh48","country_code":"HK","type":"education","lineage":["https://openalex.org/I177725633"]}],"countries":["HK"],"is_corresponding":false,"raw_author_name":"Jiaya Jia","raw_affiliation_strings":["Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong"],"raw_orcid":"https://orcid.org/0000-0002-1246-553X","affiliations":[{"raw_affiliation_string":"Department of Computer Science and Engineering, The Chinese University of Hong Kong, Hong Kong","institution_ids":["https://openalex.org/I177725633"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":2,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.9718,"has_fulltext":false,"cited_by_count":11,"citation_normalized_percentile":{"value":0.74267608,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":96,"max":98},"biblio":{"volume":"45","issue":"11","first_page":"1","last_page":"13"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9997000098228455,"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/T10502","display_name":"Advanced Memory and Neural Computing","score":0.9997000098228455,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9990000128746033,"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"}},{"id":"https://openalex.org/T11605","display_name":"Visual Attention and Saliency Detection","score":0.9973999857902527,"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":0.7544277906417847},{"id":"https://openalex.org/keywords/transformer","display_name":"Transformer","score":0.6897333264350891},{"id":"https://openalex.org/keywords/pairwise-comparison","display_name":"Pairwise comparison","score":0.5569608807563782},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.528231143951416},{"id":"https://openalex.org/keywords/security-token","display_name":"Security token","score":0.49696019291877747},{"id":"https://openalex.org/keywords/architecture","display_name":"Architecture","score":0.49538886547088623},{"id":"https://openalex.org/keywords/machine-learning","display_name":"Machine learning","score":0.36999595165252686},{"id":"https://openalex.org/keywords/computer-engineering","display_name":"Computer engineering","score":0.34698566794395447},{"id":"https://openalex.org/keywords/engineering","display_name":"Engineering","score":0.12885481119155884},{"id":"https://openalex.org/keywords/voltage","display_name":"Voltage","score":0.12840956449508667},{"id":"https://openalex.org/keywords/computer-security","display_name":"Computer security","score":0.12522581219673157},{"id":"https://openalex.org/keywords/electrical-engineering","display_name":"Electrical engineering","score":0.10901495814323425}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.7544277906417847},{"id":"https://openalex.org/C66322947","wikidata":"https://www.wikidata.org/wiki/Q11658","display_name":"Transformer","level":3,"score":0.6897333264350891},{"id":"https://openalex.org/C184898388","wikidata":"https://www.wikidata.org/wiki/Q1435712","display_name":"Pairwise comparison","level":2,"score":0.5569608807563782},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.528231143951416},{"id":"https://openalex.org/C48145219","wikidata":"https://www.wikidata.org/wiki/Q1335365","display_name":"Security token","level":2,"score":0.49696019291877747},{"id":"https://openalex.org/C123657996","wikidata":"https://www.wikidata.org/wiki/Q12271","display_name":"Architecture","level":2,"score":0.49538886547088623},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.36999595165252686},{"id":"https://openalex.org/C113775141","wikidata":"https://www.wikidata.org/wiki/Q428691","display_name":"Computer engineering","level":1,"score":0.34698566794395447},{"id":"https://openalex.org/C127413603","wikidata":"https://www.wikidata.org/wiki/Q11023","display_name":"Engineering","level":0,"score":0.12885481119155884},{"id":"https://openalex.org/C165801399","wikidata":"https://www.wikidata.org/wiki/Q25428","display_name":"Voltage","level":2,"score":0.12840956449508667},{"id":"https://openalex.org/C38652104","wikidata":"https://www.wikidata.org/wiki/Q3510521","display_name":"Computer security","level":1,"score":0.12522581219673157},{"id":"https://openalex.org/C119599485","wikidata":"https://www.wikidata.org/wiki/Q43035","display_name":"Electrical engineering","level":1,"score":0.10901495814323425},{"id":"https://openalex.org/C153349607","wikidata":"https://www.wikidata.org/wiki/Q36649","display_name":"Visual arts","level":1,"score":0.0},{"id":"https://openalex.org/C142362112","wikidata":"https://www.wikidata.org/wiki/Q735","display_name":"Art","level":0,"score":0.0}],"mesh":[],"locations_count":3,"locations":[{"id":"doi:10.1109/tpami.2022.3231442","is_oa":false,"landing_page_url":"https://doi.org/10.1109/tpami.2022.3231442","pdf_url":null,"source":{"id":"https://openalex.org/S199944782","display_name":"IEEE Transactions on Pattern Analysis and Machine Intelligence","issn_l":"0162-8828","issn":["0162-8828","1939-3539","2160-9292"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310320439","host_organization_name":"IEEE Computer Society","host_organization_lineage":["https://openalex.org/P4310320439","https://openalex.org/P4310319808"],"host_organization_lineage_names":["IEEE Computer Society","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 Transactions on Pattern Analysis and Machine Intelligence","raw_type":"journal-article"},{"id":"pmid:37015534","is_oa":false,"landing_page_url":"https://pubmed.ncbi.nlm.nih.gov/37015534","pdf_url":null,"source":{"id":"https://openalex.org/S4306525036","display_name":"PubMed","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I1299303238","host_organization_name":"National Institutes of Health","host_organization_lineage":["https://openalex.org/I1299303238"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"IEEE transactions on pattern analysis and machine intelligence","raw_type":null},{"id":"pmh:oai:repository.hkust.edu.hk:1783.1-140760","is_oa":false,"landing_page_url":"http://repository.hkust.edu.hk/ir/Record/1783.1-140760","pdf_url":null,"source":{"id":"https://openalex.org/S4306401796","display_name":"Rare & Special e-Zone (The Hong Kong University of Science and Technology)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I200769079","host_organization_name":"Hong Kong University of Science and Technology","host_organization_lineage":["https://openalex.org/I200769079"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"Article"}],"best_oa_location":null,"sustainable_development_goals":[{"display_name":"Industry, innovation and infrastructure","id":"https://metadata.un.org/sdg/9","score":0.4300000071525574}],"awards":[],"funders":[],"has_content":{"pdf":false,"grobid_xml":false},"content_urls":null,"referenced_works_count":128,"referenced_works":["https://openalex.org/W1959608418","https://openalex.org/W2108598243","https://openalex.org/W2138204974","https://openalex.org/W2299668505","https://openalex.org/W2331128040","https://openalex.org/W2502312327","https://openalex.org/W2507296351","https://openalex.org/W2603777577","https://openalex.org/W2768282280","https://openalex.org/W2787579267","https://openalex.org/W2803832867","https://openalex.org/W2883725317","https://openalex.org/W2884366600","https://openalex.org/W2884822772","https://openalex.org/W2887997457","https://openalex.org/W2890560993","https://openalex.org/W2896457183","https://openalex.org/W2962784628","https://openalex.org/W2962793481","https://openalex.org/W2962974533","https://openalex.org/W2963047948","https://openalex.org/W2963150697","https://openalex.org/W2963351448","https://openalex.org/W2963865469","https://openalex.org/W2963979492","https://openalex.org/W2964241181","https://openalex.org/W2964309429","https://openalex.org/W2970820321","https://openalex.org/W2980360762","https://openalex.org/W2996320484","https://openalex.org/W2997574889","https://openalex.org/W3030520226","https://openalex.org/W3034772996","https://openalex.org/W3034973631","https://openalex.org/W3035682985","https://openalex.org/W3092462694","https://openalex.org/W3093960091","https://openalex.org/W3094502228","https://openalex.org/W3109301572","https://openalex.org/W3115462295","https://openalex.org/W3121523901","https://openalex.org/W3125127048","https://openalex.org/W3131500599","https://openalex.org/W3133696297","https://openalex.org/W3138516171","https://openalex.org/W3139049060","https://openalex.org/W3151130473","https://openalex.org/W3157528469","https://openalex.org/W3168124404","https://openalex.org/W3170188883","https://openalex.org/W3170841864","https://openalex.org/W3170863103","https://openalex.org/W3170874841","https://openalex.org/W3171087525","https://openalex.org/W3172801447","https://openalex.org/W3174402370","https://openalex.org/W3174418826","https://openalex.org/W3175515048","https://openalex.org/W3175544090","https://openalex.org/W3176153963","https://openalex.org/W3179869055","https://openalex.org/W3187216907","https://openalex.org/W3203020760","https://openalex.org/W3210646981","https://openalex.org/W3211490618","https://openalex.org/W3213122141","https://openalex.org/W3214897340","https://openalex.org/W4214493665","https://openalex.org/W4214614183","https://openalex.org/W4214634256","https://openalex.org/W4214669216","https://openalex.org/W4214709605","https://openalex.org/W4214713996","https://openalex.org/W4287062814","https://openalex.org/W4287115696","https://openalex.org/W4288325606","https://openalex.org/W4295727797","https://openalex.org/W4313156423","https://openalex.org/W4322716158","https://openalex.org/W4385245566","https://openalex.org/W6640963894","https://openalex.org/W6680106237","https://openalex.org/W6681096077","https://openalex.org/W6697673422","https://openalex.org/W6724670942","https://openalex.org/W6724804524","https://openalex.org/W6739901393","https://openalex.org/W6746052068","https://openalex.org/W6748655329","https://openalex.org/W6751795773","https://openalex.org/W6752051073","https://openalex.org/W6753421600","https://openalex.org/W6754278344","https://openalex.org/W6754723338","https://openalex.org/W6755207826","https://openalex.org/W6757376125","https://openalex.org/W6764322716","https://openalex.org/W6768371451","https://openalex.org/W6768841368","https://openalex.org/W6772381481","https://openalex.org/W6778485988","https://openalex.org/W6780226713","https://openalex.org/W6780724485","https://openalex.org/W6784094891","https://openalex.org/W6784333009","https://openalex.org/W6788135285","https://openalex.org/W6789802742","https://openalex.org/W6790690058","https://openalex.org/W6792695861","https://openalex.org/W6794345597","https://openalex.org/W6796321339","https://openalex.org/W6796494063","https://openalex.org/W6796580215","https://openalex.org/W6796716886","https://openalex.org/W6796721132","https://openalex.org/W6796761347","https://openalex.org/W6796931752","https://openalex.org/W6797019122","https://openalex.org/W6797399245","https://openalex.org/W6797533734","https://openalex.org/W6797790494","https://openalex.org/W6798016242","https://openalex.org/W6802855191","https://openalex.org/W6803028189","https://openalex.org/W6803508786","https://openalex.org/W6803870738","https://openalex.org/W6841287799","https://openalex.org/W6892006130"],"related_works":["https://openalex.org/W2487162673","https://openalex.org/W2942366970","https://openalex.org/W2793211469","https://openalex.org/W2949152769","https://openalex.org/W4372354731","https://openalex.org/W2562400057","https://openalex.org/W1692008701","https://openalex.org/W4388335561","https://openalex.org/W2597588799","https://openalex.org/W3016124757"],"abstract_inverted_index":{"The":[0,94],"architecture":[1],"of":[2,33,46,133,148,152],"transformers,":[3,47],"which":[4],"recently":[5],"witness":[6],"booming":[7],"applications":[8],"in":[9,59,113,154],"vision":[10,92,155],"tasks,":[11],"has":[12,54],"pivoted":[13],"against":[14],"the":[15,21,42,114,146,149],"widespread":[16],"convolutional":[17],"paradigm.":[18],"Relying":[19],"on":[20,121],"tokenization":[22],"process":[23],"that":[24],"splits":[25],"inputs":[26],"into":[27],"multiple":[28],"tokens,":[29],"transformers":[30],"are":[31],"capable":[32],"extracting":[34],"their":[35],"pairwise":[36],"relationships":[37],"using":[38],"self-attention.":[39],"While":[40],"being":[41],"stemming":[43],"building":[44],"block":[45],"what":[48],"makes":[49],"for":[50,91],"a":[51,107],"good":[52],"tokenizer":[53],"not":[55],"been":[56],"well":[57],"understood":[58],"computer":[60],"vision.":[61],"In":[62,75],"this":[63,67],"work,":[64],"we":[65,125],"investigate":[66],"uncharted":[68],"problem":[69],"from":[70],"an":[71],"information":[72],"trade-off":[73],"perspective.":[74],"addition":[76],"to":[77,87],"unifying":[78],"and":[79,130],"understanding":[80],"existing":[81],"structural":[82],"modifications,":[83],"our":[84],"derivation":[85],"leads":[86],"better":[88],"design":[89],"strategies":[90],"tokenizers.":[93],"proposed":[95],"Modulation":[96],"across":[97],"Tokens":[98],"(MoTo)":[99],"incorporates":[100],"inter-token":[101],"modeling":[102],"capability":[103],"through":[104],"normalization.":[105],"Furthermore,":[106],"regularization":[108],"objective":[109],"TokenProp":[110],"is":[111],"embraced":[112],"standard":[115],"training":[116],"regime.":[117],"Through":[118],"extensive":[119],"experiments":[120],"various":[122],"transformer":[123],"architectures,":[124],"observe":[126],"both":[127],"improved":[128],"performance":[129],"intriguing":[131],"properties":[132],"these":[134],"two":[135],"plug-and-play":[136],"designs":[137,151],"with":[138],"negligible":[139],"computational":[140],"overhead.":[141],"These":[142],"observations":[143],"further":[144],"indicate":[145],"importance":[147],"commonly-omitted":[150],"tokenizers":[153],"transformer.":[156]},"counts_by_year":[{"year":2025,"cited_by_count":4},{"year":2024,"cited_by_count":4},{"year":2023,"cited_by_count":3}],"updated_date":"2026-03-27T05:58:40.876381","created_date":"2025-10-10T00:00:00"}