{"id":"https://openalex.org/W4390871684","doi":"https://doi.org/10.1109/iccv51070.2023.01557","title":"Jumping through Local Minima: Quantization in the Loss Landscape of Vision Transformers","display_name":"Jumping through Local Minima: Quantization in the Loss Landscape of Vision Transformers","publication_year":2023,"publication_date":"2023-10-01","ids":{"openalex":"https://openalex.org/W4390871684","doi":"https://doi.org/10.1109/iccv51070.2023.01557"},"language":"en","primary_location":{"id":"doi:10.1109/iccv51070.2023.01557","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iccv51070.2023.01557","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2023 IEEE/CVF International Conference on Computer Vision (ICCV)","raw_type":"proceedings-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/A5058909899","display_name":"Natalia Frumkin","orcid":null},"institutions":[{"id":"https://openalex.org/I86519309","display_name":"The University of Texas at Austin","ror":"https://ror.org/00hj54h04","country_code":"US","type":"education","lineage":["https://openalex.org/I86519309"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Natalia Frumkin","raw_affiliation_strings":["The University of Texas at Austin"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Texas at Austin","institution_ids":["https://openalex.org/I86519309"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5078130628","display_name":"Dibakar Gope","orcid":null},"institutions":[{"id":"https://openalex.org/I4210156213","display_name":"American Rock Mechanics Association","ror":"https://ror.org/05vfrxy92","country_code":"US","type":"nonprofit","lineage":["https://openalex.org/I4210156213"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Dibakar Gope","raw_affiliation_strings":["Arm Inc"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Arm Inc","institution_ids":["https://openalex.org/I4210156213"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5065985595","display_name":"Diana Marculescu","orcid":"https://orcid.org/0000-0002-5734-4221"},"institutions":[{"id":"https://openalex.org/I86519309","display_name":"The University of Texas at Austin","ror":"https://ror.org/00hj54h04","country_code":"US","type":"education","lineage":["https://openalex.org/I86519309"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Diana Marculescu","raw_affiliation_strings":["The University of Texas at Austin"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"The University of Texas at Austin","institution_ids":["https://openalex.org/I86519309"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":1.2351,"has_fulltext":false,"cited_by_count":11,"citation_normalized_percentile":{"value":0.82388044,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":96,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"16932","last_page":"16942"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9998999834060669,"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"}},"topics":[{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9998999834060669,"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/T11992","display_name":"CCD and CMOS Imaging Sensors","score":0.9973999857902527,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.9965000152587891,"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/quantization","display_name":"Quantization (signal processing)","score":0.7127265930175781},{"id":"https://openalex.org/keywords/maxima-and-minima","display_name":"Maxima and minima","score":0.6525735855102539},{"id":"https://openalex.org/keywords/traverse","display_name":"Traverse","score":0.6121987104415894},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.563194751739502},{"id":"https://openalex.org/keywords/streak","display_name":"Streak","score":0.4882793426513672},{"id":"https://openalex.org/keywords/overfitting","display_name":"Overfitting","score":0.4850101172924042},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.42715516686439514},{"id":"https://openalex.org/keywords/robustness","display_name":"Robustness (evolution)","score":0.42252442240715027},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.40899211168289185},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.3498344421386719},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.31359463930130005},{"id":"https://openalex.org/keywords/physics","display_name":"Physics","score":0.14592492580413818},{"id":"https://openalex.org/keywords/cartography","display_name":"Cartography","score":0.11185809969902039},{"id":"https://openalex.org/keywords/biology","display_name":"Biology","score":0.08503630757331848},{"id":"https://openalex.org/keywords/mathematical-analysis","display_name":"Mathematical analysis","score":0.08226335048675537},{"id":"https://openalex.org/keywords/optics","display_name":"Optics","score":0.07837089896202087}],"concepts":[{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.7127265930175781},{"id":"https://openalex.org/C186633575","wikidata":"https://www.wikidata.org/wiki/Q845060","display_name":"Maxima and minima","level":2,"score":0.6525735855102539},{"id":"https://openalex.org/C176809094","wikidata":"https://www.wikidata.org/wiki/Q15401496","display_name":"Traverse","level":2,"score":0.6121987104415894},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.563194751739502},{"id":"https://openalex.org/C65185188","wikidata":"https://www.wikidata.org/wiki/Q107775","display_name":"Streak","level":2,"score":0.4882793426513672},{"id":"https://openalex.org/C22019652","wikidata":"https://www.wikidata.org/wiki/Q331309","display_name":"Overfitting","level":3,"score":0.4850101172924042},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.42715516686439514},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.42252442240715027},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.40899211168289185},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.3498344421386719},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.31359463930130005},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.14592492580413818},{"id":"https://openalex.org/C58640448","wikidata":"https://www.wikidata.org/wiki/Q42515","display_name":"Cartography","level":1,"score":0.11185809969902039},{"id":"https://openalex.org/C86803240","wikidata":"https://www.wikidata.org/wiki/Q420","display_name":"Biology","level":0,"score":0.08503630757331848},{"id":"https://openalex.org/C134306372","wikidata":"https://www.wikidata.org/wiki/Q7754","display_name":"Mathematical analysis","level":1,"score":0.08226335048675537},{"id":"https://openalex.org/C120665830","wikidata":"https://www.wikidata.org/wiki/Q14620","display_name":"Optics","level":1,"score":0.07837089896202087},{"id":"https://openalex.org/C205649164","wikidata":"https://www.wikidata.org/wiki/Q1071","display_name":"Geography","level":0,"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/C104317684","wikidata":"https://www.wikidata.org/wiki/Q7187","display_name":"Gene","level":2,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/iccv51070.2023.01557","is_oa":false,"landing_page_url":"https://doi.org/10.1109/iccv51070.2023.01557","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2023 IEEE/CVF International Conference on Computer Vision (ICCV)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":33,"referenced_works":["https://openalex.org/W2626696598","https://openalex.org/W2774185825","https://openalex.org/W2903972532","https://openalex.org/W2963122961","https://openalex.org/W2970601456","https://openalex.org/W2998218113","https://openalex.org/W3034997257","https://openalex.org/W3035060554","https://openalex.org/W3038470071","https://openalex.org/W3094502228","https://openalex.org/W3138516171","https://openalex.org/W3145450063","https://openalex.org/W3166859509","https://openalex.org/W3173374050","https://openalex.org/W3202442802","https://openalex.org/W3211787299","https://openalex.org/W4221155876","https://openalex.org/W4226174177","https://openalex.org/W4226339979","https://openalex.org/W4285601701","https://openalex.org/W4293023518","https://openalex.org/W4297808394","https://openalex.org/W4309125142","https://openalex.org/W4312376588","https://openalex.org/W4312757094","https://openalex.org/W4313069943","https://openalex.org/W4321637298","https://openalex.org/W4390873361","https://openalex.org/W6767298317","https://openalex.org/W6776767859","https://openalex.org/W6779326418","https://openalex.org/W6795991173","https://openalex.org/W6797854001"],"related_works":["https://openalex.org/W2377402383","https://openalex.org/W2380835401","https://openalex.org/W4362597605","https://openalex.org/W2381912691","https://openalex.org/W2350381577","https://openalex.org/W2353618196","https://openalex.org/W1574414179","https://openalex.org/W2348074676","https://openalex.org/W2385033175","https://openalex.org/W2374043190"],"abstract_inverted_index":{"Quantization":[0],"scale":[1,59],"and":[2,150,155,167],"bit-width":[3],"are":[4],"the":[5,103,120,139],"most":[6],"important":[7],"parameters":[8],"when":[9,37],"considering":[10],"how":[11],"to":[12,40,100],"quantize":[13],"a":[14,25,45,65,132,143,163],"neural":[15],"network.":[16],"Prior":[17],"work":[18],"focuses":[19],"on":[20,119,162],"optimizing":[21],"quantization":[22,41,58,158,176],"scales":[23],"in":[24,57,71,174],"global":[26],"manner":[27],"through":[28],"gradient":[29,80],"methods":[30,81],"(gradient":[31],"descent":[32],"&":[33],"Hessian":[34],"analysis).":[35],"Yet,":[36],"applying":[38],"perturbations":[39,56],"scales,":[42],"we":[43,96,107],"observe":[44],"very":[46],"jagged,":[47],"highly":[48,133],"non-smooth":[49,104,134],"test":[50],"loss":[51],"landscape.":[52,105],"In":[53,77,91],"fact,":[54],"small":[55,121],"can":[60],"greatly":[61],"affect":[62],"accuracy,":[63],"yielding":[64],"0.5":[66],"\u2013":[67],"0.8%":[68],"accuracy":[69,141],"boost":[70],"4-bit":[72],"quantized":[73,145],"vision":[74],"transformers":[75],"(ViTs).":[76],"this":[78],"regime,":[79],"break":[82],"down,":[83],"since":[84],"they":[85],"cannot":[86],"reliably":[87],"reach":[88],"local":[89],"minima.":[90],"our":[92],"work,":[93],"dubbed":[94],"Evol-Q,":[95],"use":[97],"evolutionary":[98],"search":[99],"effectively":[101],"traverse":[102],"Additionally,":[106],"propose":[108],"using":[109],"an":[110],"infoNCE":[111],"loss,":[112],"which":[113],"not":[114],"only":[115],"helps":[116],"combat":[117],"overfitting":[118],"calibration":[122],"dataset":[123],"(1,":[124],"000":[125],"images)":[126],"but":[127],"also":[128],"makes":[129],"traversing":[130],"such":[131],"surface":[135],"easier.":[136],"Evol-Q":[137],"improves":[138],"top-1":[140],"of":[142,165],"fully":[144],"ViT-Base":[146],"by":[147],"10.30%,":[148],"0.78%,":[149],"0.15%":[151],"for":[152],"3-bit,":[153],"4-bit,":[154],"8-bit":[156],"weight":[157],"levels.":[159],"Extensive":[160],"experiments":[161],"variety":[164],"CNN":[166],"ViT":[168],"architectures":[169],"further":[170],"demonstrate":[171],"its":[172],"robustness":[173],"extreme":[175],"scenarios.":[177],"Our":[178],"code":[179],"is":[180],"available":[181],"at":[182],"https:":[183],"//":[184],"github.com/enyac-group/evol-q":[185],".":[186]},"counts_by_year":[{"year":2026,"cited_by_count":3},{"year":2025,"cited_by_count":5},{"year":2024,"cited_by_count":3}],"updated_date":"2026-06-11T09:08:48.828518","created_date":"2025-10-10T00:00:00"}