{"id":"https://openalex.org/W4415538525","doi":"https://doi.org/10.1145/3746027.3755856","title":"Quantization Meets OOD: Generalizable Quantization-aware Training from a Flatness Perspective","display_name":"Quantization Meets OOD: Generalizable Quantization-aware Training from a Flatness Perspective","publication_year":2025,"publication_date":"2025-10-25","ids":{"openalex":"https://openalex.org/W4415538525","doi":"https://doi.org/10.1145/3746027.3755856"},"language":"en","primary_location":{"id":"doi:10.1145/3746027.3755856","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3746027.3755856","pdf_url":null,"source":null,"license":"cc-by-nc-sa","license_id":"https://openalex.org/licenses/cc-by-nc-sa","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd ACM International Conference on Multimedia","raw_type":"proceedings-article"},"type":"preprint","indexed_in":["arxiv","crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://doi.org/10.1145/3746027.3755856","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":null,"display_name":"Jiacheng Jiang","orcid":"https://orcid.org/0009-0006-5791-0664"},"institutions":[{"id":"https://openalex.org/I3131625388","display_name":"University Town of Shenzhen","ror":"https://ror.org/05f5j6225","country_code":"CN","type":"education","lineage":["https://openalex.org/I3131625388"]},{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Jiacheng Jiang","raw_affiliation_strings":["SIGS, Tsinghua University, Shenzhen, China"],"affiliations":[{"raw_affiliation_string":"SIGS, Tsinghua University, Shenzhen, China","institution_ids":["https://openalex.org/I3131625388","https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5103072786","display_name":"Yuan Meng","orcid":"https://orcid.org/0000-0002-7450-9438"},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Yuan Meng","raw_affiliation_strings":["Key Laboratory of Pervasive Computing, Ministry of Education Department of Computer Science and Technology, Tsinghua University, Beijing, China and Department of Computer Science and Technology, Tsinghua University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"Key Laboratory of Pervasive Computing, Ministry of Education Department of Computer Science and Technology, Tsinghua University, Beijing, China and Department of Computer Science and Technology, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5101955396","display_name":"Chen Tang","orcid":"https://orcid.org/0000-0002-0108-6729"},"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":"Chen Tang","raw_affiliation_strings":["MMLab, The Chinese University of Hong Kong, Hong Kong, China"],"affiliations":[{"raw_affiliation_string":"MMLab, The Chinese University of Hong Kong, Hong Kong, China","institution_ids":["https://openalex.org/I177725633"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5102516031","display_name":"Han Yu","orcid":"https://orcid.org/0009-0000-2828-4541"},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Han Yu","raw_affiliation_strings":["Department of Computer Science and Technology, Tsinghua University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"Department of Computer Science and Technology, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":null,"display_name":"Qun Li","orcid":"https://orcid.org/0009-0002-7448-6261"},"institutions":[{"id":"https://openalex.org/I3131625388","display_name":"University Town of Shenzhen","ror":"https://ror.org/05f5j6225","country_code":"CN","type":"education","lineage":["https://openalex.org/I3131625388"]},{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Qun Li","raw_affiliation_strings":["SIGS, Tsinghua University, Shenzhen, China"],"affiliations":[{"raw_affiliation_string":"SIGS, Tsinghua University, Shenzhen, China","institution_ids":["https://openalex.org/I3131625388","https://openalex.org/I99065089"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100376399","display_name":"Zhi Wang","orcid":"https://orcid.org/0000-0002-5462-6178"},"institutions":[{"id":"https://openalex.org/I3131625388","display_name":"University Town of Shenzhen","ror":"https://ror.org/05f5j6225","country_code":"CN","type":"education","lineage":["https://openalex.org/I3131625388"]},{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zhi Wang","raw_affiliation_strings":["SIGS, Tsinghua University, Shenzhen, China"],"affiliations":[{"raw_affiliation_string":"SIGS, Tsinghua University, Shenzhen, China","institution_ids":["https://openalex.org/I3131625388","https://openalex.org/I99065089"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100339293","display_name":"Wenwu Zhu","orcid":"https://orcid.org/0000-0003-2236-9290"},"institutions":[{"id":"https://openalex.org/I99065089","display_name":"Tsinghua University","ror":"https://ror.org/03cve4549","country_code":"CN","type":"education","lineage":["https://openalex.org/I99065089"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Wenwu Zhu","raw_affiliation_strings":["Department of Computer Science and Technology, Tsinghua University, Beijing, China and Key Laboratory of Pervasive Computing, Ministry of Education Department of Computer Science and Technology, Tsinghua University, Beijing, China"],"affiliations":[{"raw_affiliation_string":"Department of Computer Science and Technology, Tsinghua University, Beijing, China and Key Laboratory of Pervasive Computing, Ministry of Education Department of Computer Science and Technology, Tsinghua University, Beijing, China","institution_ids":["https://openalex.org/I99065089"]}]}],"institutions":[],"countries_distinct_count":2,"institutions_distinct_count":7,"corresponding_author_ids":[],"corresponding_institution_ids":["https://openalex.org/I3131625388","https://openalex.org/I99065089"],"apc_list":null,"apc_paid":null,"fwci":2.3568,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.91573262,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":96,"max":98},"biblio":{"volume":null,"issue":null,"first_page":"11937","last_page":"11946"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.9915000200271606,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.9915000200271606,"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/T10320","display_name":"Neural Networks and Applications","score":0.9865000247955322,"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/T12676","display_name":"Machine Learning and ELM","score":0.9860000014305115,"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/flatness","display_name":"Flatness (cosmology)","score":0.5235000252723694},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.48739999532699585},{"id":"https://openalex.org/keywords/generalization","display_name":"Generalization","score":0.46880000829696655},{"id":"https://openalex.org/keywords/perspective","display_name":"Perspective (graphical)","score":0.4323999881744385},{"id":"https://openalex.org/keywords/focus","display_name":"Focus (optics)","score":0.3898000121116638},{"id":"https://openalex.org/keywords/benchmark","display_name":"Benchmark (surveying)","score":0.38589999079704285},{"id":"https://openalex.org/keywords/training","display_name":"Training (meteorology)","score":0.37439998984336853}],"concepts":[{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5652999877929688},{"id":"https://openalex.org/C2778530986","wikidata":"https://www.wikidata.org/wiki/Q5457948","display_name":"Flatness (cosmology)","level":3,"score":0.5235000252723694},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.5121999979019165},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.48739999532699585},{"id":"https://openalex.org/C177148314","wikidata":"https://www.wikidata.org/wiki/Q170084","display_name":"Generalization","level":2,"score":0.46880000829696655},{"id":"https://openalex.org/C12713177","wikidata":"https://www.wikidata.org/wiki/Q1900281","display_name":"Perspective (graphical)","level":2,"score":0.4323999881744385},{"id":"https://openalex.org/C192209626","wikidata":"https://www.wikidata.org/wiki/Q190909","display_name":"Focus (optics)","level":2,"score":0.3898000121116638},{"id":"https://openalex.org/C185798385","wikidata":"https://www.wikidata.org/wiki/Q1161707","display_name":"Benchmark (surveying)","level":2,"score":0.38589999079704285},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.375},{"id":"https://openalex.org/C2777211547","wikidata":"https://www.wikidata.org/wiki/Q17141490","display_name":"Training (meteorology)","level":2,"score":0.37439998984336853},{"id":"https://openalex.org/C176217482","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Metric (unit)","level":2,"score":0.3393999934196472},{"id":"https://openalex.org/C2780898871","wikidata":"https://www.wikidata.org/wiki/Q860554","display_name":"Performance metric","level":2,"score":0.33739998936653137},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.3312000036239624},{"id":"https://openalex.org/C118505674","wikidata":"https://www.wikidata.org/wiki/Q42586063","display_name":"Encoder","level":2,"score":0.29330000281333923},{"id":"https://openalex.org/C63479239","wikidata":"https://www.wikidata.org/wiki/Q7353546","display_name":"Robustness (evolution)","level":3,"score":0.28029999136924744},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.27320000529289246},{"id":"https://openalex.org/C2780069185","wikidata":"https://www.wikidata.org/wiki/Q7977945","display_name":"Equivalence (formal languages)","level":2,"score":0.2614000141620636},{"id":"https://openalex.org/C2776214188","wikidata":"https://www.wikidata.org/wiki/Q408386","display_name":"Inference","level":2,"score":0.25600001215934753},{"id":"https://openalex.org/C46312889","wikidata":"https://www.wikidata.org/wiki/Q950502","display_name":"Nanomanufacturing","level":2,"score":0.25}],"mesh":[],"locations_count":2,"locations":[{"id":"doi:10.1145/3746027.3755856","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3746027.3755856","pdf_url":null,"source":null,"license":"cc-by-nc-sa","license_id":"https://openalex.org/licenses/cc-by-nc-sa","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd ACM International Conference on Multimedia","raw_type":"proceedings-article"},{"id":"pmh:oai:arXiv.org:2509.00859","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2509.00859","pdf_url":"https://arxiv.org/pdf/2509.00859","source":{"id":"https://openalex.org/S4393918464","display_name":"ArXiv.org","issn_l":"2331-8422","issn":["2331-8422"],"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":"text"}],"best_oa_location":{"id":"doi:10.1145/3746027.3755856","is_oa":true,"landing_page_url":"https://doi.org/10.1145/3746027.3755856","pdf_url":null,"source":null,"license":"cc-by-nc-sa","license_id":"https://openalex.org/licenses/cc-by-nc-sa","version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the 33rd ACM International Conference on Multimedia","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G4326992788","display_name":null,"funder_award_id":"92467204 & 62472249 & 62402264","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":14,"referenced_works":["https://openalex.org/W2194775991","https://openalex.org/W2276486856","https://openalex.org/W2606722458","https://openalex.org/W2627183927","https://openalex.org/W2958360136","https://openalex.org/W2963043696","https://openalex.org/W2963480671","https://openalex.org/W2964124968","https://openalex.org/W3038585207","https://openalex.org/W3176946833","https://openalex.org/W4312232143","https://openalex.org/W4312391126","https://openalex.org/W4312774735","https://openalex.org/W4386066238"],"related_works":[],"abstract_inverted_index":{"Current":[0],"quantization-aware":[1],"training":[2,138],"(QAT)":[3],"methods":[4],"primarily":[5],"focus":[6],"on":[7,14,24,166],"enhancing":[8],"the":[9,20,54,57,71,83,108,142,157,171],"performance":[10,22,49],"of":[11,61,173],"quantized":[12],"models":[13],"in-distribution":[15],"(I.D)":[16],"data,":[17],"while":[18],"overlooking":[19],"potential":[21],"degradation":[23],"out-of-distribution":[25],"(OOD)":[26],"data.":[27],"In":[28],"this":[29,34],"paper,":[30],"we":[31,52,87],"first":[32],"substantiate":[33],"problem":[35],"through":[36],"rigorous":[37],"experiment,":[38],"showing":[39],"that":[40,59,73],"QAT":[41,74,91,118],"can":[42,81],"lead":[43,75],"to":[44,66,76,94,106,129,134,155],"a":[45,77,89,102],"significant":[46],"OOD":[47,68,168,183],"generalization":[48,69],"degradation.":[50],"Further,":[51],"find":[53],"contradiction":[55],"between":[56,112,147],"perspective":[58],"flatness":[60],"loss":[62,79],"landscape":[63],"gives":[64],"rise":[65],"superior":[67],"and":[70,119,182],"phenomenon":[72],"sharp":[78],"landscape,":[80],"cause":[82],"above":[84],"problem.":[85],"Therefore,":[86],"propose":[88],"flatness-oriented":[90],"method,":[92],"FQAT,":[93],"achieve":[95],"generalizable":[96],"QAT.":[97],"Specifically,":[98],"i)":[99],"FQAT":[100,122],"introduces":[101],"layer-wise":[103],"freezing":[104,127],"mechanism":[105],"mitigate":[107],"gradient":[109,150],"conflict":[110],"issue":[111],"dual":[113],"optimization":[114],"objectives":[115],"(i.e.,":[116],"vanilla":[117],"flatness).":[120],"ii)":[121],"proposes":[123],"an":[124],"disorder-guided":[125],"adaptive":[126],"algorithm":[128,158],"dynamically":[130],"determines":[131],"which":[132],"layers":[133,161],"freeze":[135],"at":[136],"each":[137],"step,":[139],"effectively":[140],"addressing":[141],"challenges":[143],"caused":[144],"by":[145],"interference":[146],"layers.":[148],"A":[149],"disorder":[151],"metric":[152],"is":[153],"designed":[154],"help":[156],"identify":[159],"unstable":[160],"during":[162],"training.":[163],"Extensive":[164],"experiments":[165],"influential":[167],"benchmark":[169],"demonstrate":[170],"superiority":[172],"our":[174],"method":[175],"over":[176],"state-of-the-art":[177],"baselines":[178],"under":[179],"both":[180],"I.D":[181],"image":[184],"classification":[185],"tasks.":[186]},"counts_by_year":[{"year":2026,"cited_by_count":1}],"updated_date":"2026-04-09T08:11:56.329763","created_date":"2025-10-25T00:00:00"}