{"id":"https://openalex.org/W4386615681","doi":"https://doi.org/10.1109/icip49359.2023.10222516","title":"Designing Strong Baselines for Ternary Neural Network Quantization through Support and Mass Equalization","display_name":"Designing Strong Baselines for Ternary Neural Network Quantization through Support and Mass Equalization","publication_year":2023,"publication_date":"2023-09-11","ids":{"openalex":"https://openalex.org/W4386615681","doi":"https://doi.org/10.1109/icip49359.2023.10222516"},"language":"en","primary_location":{"id":"doi:10.1109/icip49359.2023.10222516","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/icip49359.2023.10222516","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2023 IEEE International Conference on Image Processing (ICIP)","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/A5085696448","display_name":"Edouard Yvinec","orcid":"https://orcid.org/0000-0002-4318-612X"},"institutions":[{"id":"https://openalex.org/I1294671590","display_name":"Centre National de la Recherche Scientifique","ror":"https://ror.org/02feahw73","country_code":"FR","type":"government","lineage":["https://openalex.org/I1294671590"]},{"id":"https://openalex.org/I39804081","display_name":"Sorbonne Universit\u00e9","ror":"https://ror.org/02en5vm52","country_code":"FR","type":"education","lineage":["https://openalex.org/I39804081"]}],"countries":["FR"],"is_corresponding":false,"raw_author_name":"Edouard Yvinec","raw_affiliation_strings":["Sorbonne Université, CNRS, ISIR,Paris,France,75005"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Sorbonne Université, CNRS, ISIR,Paris,France,75005","institution_ids":["https://openalex.org/I39804081","https://openalex.org/I1294671590"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5037619048","display_name":"Arnaud Dapogny","orcid":"https://orcid.org/0000-0002-0074-8719"},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Arnaud Dapogny","raw_affiliation_strings":["Datakalab,Paris,France,75017"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Datakalab,Paris,France,75017","institution_ids":[]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5022871131","display_name":"K\u00e9vin Bailly","orcid":"https://orcid.org/0000-0001-7802-3673"},"institutions":[{"id":"https://openalex.org/I1294671590","display_name":"Centre National de la Recherche Scientifique","ror":"https://ror.org/02feahw73","country_code":"FR","type":"government","lineage":["https://openalex.org/I1294671590"]},{"id":"https://openalex.org/I39804081","display_name":"Sorbonne Universit\u00e9","ror":"https://ror.org/02en5vm52","country_code":"FR","type":"education","lineage":["https://openalex.org/I39804081"]}],"countries":["FR"],"is_corresponding":false,"raw_author_name":"Kevin Bailly","raw_affiliation_strings":["Sorbonne Université, CNRS, ISIR,Paris,France,75005"],"raw_orcid":null,"affiliations":[{"raw_affiliation_string":"Sorbonne Université, CNRS, ISIR,Paris,France,75005","institution_ids":["https://openalex.org/I39804081","https://openalex.org/I1294671590"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":3,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":0,"citation_normalized_percentile":{"value":0.10408118,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":null,"biblio":{"volume":"32","issue":null,"first_page":"540","last_page":"544"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9998000264167786,"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":0.9998000264167786,"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/T10627","display_name":"Advanced Image and Video Retrieval Techniques","score":0.9984999895095825,"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/T11307","display_name":"Domain Adaptation and Few-Shot Learning","score":0.9976000189781189,"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":0.7340224981307983},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.6375325322151184},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.6034350991249084},{"id":"https://openalex.org/keywords/dither","display_name":"Dither","score":0.5642949938774109},{"id":"https://openalex.org/keywords/ternary-operation","display_name":"Ternary operation","score":0.5062081217765808},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.49108362197875977},{"id":"https://openalex.org/keywords/minification","display_name":"Minification","score":0.43004122376441956},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.3240388035774231},{"id":"https://openalex.org/keywords/noise-shaping","display_name":"Noise shaping","score":0.13897624611854553}],"concepts":[{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.7340224981307983},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.6375325322151184},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.6034350991249084},{"id":"https://openalex.org/C70451592","wikidata":"https://www.wikidata.org/wiki/Q376493","display_name":"Dither","level":3,"score":0.5642949938774109},{"id":"https://openalex.org/C64452783","wikidata":"https://www.wikidata.org/wiki/Q1524945","display_name":"Ternary operation","level":2,"score":0.5062081217765808},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.49108362197875977},{"id":"https://openalex.org/C147764199","wikidata":"https://www.wikidata.org/wiki/Q6865248","display_name":"Minification","level":2,"score":0.43004122376441956},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3240388035774231},{"id":"https://openalex.org/C9083635","wikidata":"https://www.wikidata.org/wiki/Q2133535","display_name":"Noise shaping","level":2,"score":0.13897624611854553},{"id":"https://openalex.org/C31972630","wikidata":"https://www.wikidata.org/wiki/Q844240","display_name":"Computer vision","level":1,"score":0.0},{"id":"https://openalex.org/C199360897","wikidata":"https://www.wikidata.org/wiki/Q9143","display_name":"Programming language","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/icip49359.2023.10222516","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1109/icip49359.2023.10222516","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2023 IEEE International Conference on Image Processing (ICIP)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G3584957902","display_name":"Face Interpretation with deep and ensemble Learning","funder_award_id":"ANR-17-CE33-0002","funder_id":"https://openalex.org/F4320320883","funder_display_name":"Agence Nationale de la Recherche"}],"funders":[{"id":"https://openalex.org/F4320320883","display_name":"Agence Nationale de la Recherche","ror":"https://ror.org/00rbzpz17"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":43,"referenced_works":["https://openalex.org/W1991567646","https://openalex.org/W2108598243","https://openalex.org/W2186094539","https://openalex.org/W2194775991","https://openalex.org/W2469490737","https://openalex.org/W2560017826","https://openalex.org/W2785916643","https://openalex.org/W2787513570","https://openalex.org/W2790621718","https://openalex.org/W2955425717","https://openalex.org/W2962761403","https://openalex.org/W2963082137","https://openalex.org/W2963122961","https://openalex.org/W2963163009","https://openalex.org/W2970601456","https://openalex.org/W2981751377","https://openalex.org/W3102633758","https://openalex.org/W3166859509","https://openalex.org/W3203273637","https://openalex.org/W4221160944","https://openalex.org/W4221162983","https://openalex.org/W4288601266","https://openalex.org/W4289438390","https://openalex.org/W4300081896","https://openalex.org/W4300479382","https://openalex.org/W4317934857","https://openalex.org/W4318142560","https://openalex.org/W4375869345","https://openalex.org/W6686583229","https://openalex.org/W6720242923","https://openalex.org/W6730047919","https://openalex.org/W6733877748","https://openalex.org/W6748319235","https://openalex.org/W6748408302","https://openalex.org/W6754887999","https://openalex.org/W6758046424","https://openalex.org/W6758082782","https://openalex.org/W6762718338","https://openalex.org/W6767298317","https://openalex.org/W6776767859","https://openalex.org/W6789240164","https://openalex.org/W6809919171","https://openalex.org/W6849329879"],"related_works":["https://openalex.org/W2389128124","https://openalex.org/W2364812720","https://openalex.org/W2138725603","https://openalex.org/W1984486060","https://openalex.org/W2050335563","https://openalex.org/W2094272778","https://openalex.org/W2000868246","https://openalex.org/W4210717739","https://openalex.org/W2933484786","https://openalex.org/W2067936044"],"abstract_inverted_index":{"Deep":[0],"neural":[1,169],"networks":[2],"(DNNs)":[3],"offer":[4],"the":[5,73,86,91,102,108,141,162],"highest":[6,109],"performance":[7,142],"in":[8,14,61,151,167],"a":[9,77,147],"wide":[10],"range":[11],"of":[12,90,143,149],"applications":[13],"computer":[15],"vision.":[16],"These":[17],"results":[18],"rely":[19],"on":[20],"over-parameterized":[21],"backbones,":[22],"which":[23,94],"are":[24],"expensive":[25],"to":[26,51,70,126,138,160],"run.":[27],"This":[28,100],"computational":[29],"burden":[30],"can":[31],"be":[32],"dramatically":[33],"reduced":[34],"by":[35],"quantizing":[36],"(in":[37],"either":[38],"data-free":[39],"(DFQ),":[40],"post-training":[41],"(PTQ)":[42],"or":[43,110],"quantization-aware":[44],"training":[45],"(QAT)":[46],"scenarios)":[47],"floating":[48],"point":[49],"values":[50,53],"ternary":[52,97,144],"(2":[54],"bits,":[55],"with":[56],"each":[57],"weight":[58,92],"taking":[59],"value":[60],"{\u22121,0,1}).":[62],"In":[63],"this":[64],"context,":[65],"we":[66,117],"observe":[67],"that":[68,124,134],"rounding":[69],"nearest":[71],"minimizes":[72],"expected":[74],"error":[75],"given":[76],"uniform":[78],"distribution":[79],"and":[80,88,122,154,156],"thus":[81],"does":[82],"not":[83],"account":[84],"for":[85,164],"skewness":[87],"kurtosis":[89],"distribution,":[93],"strongly":[95],"affects":[96],"quantization":[98,112,145],"performance.":[99],"raises":[101],"following":[103],"question:":[104],"shall":[105],"one":[106],"minimize":[107],"average":[111],"error?":[113],"To":[114],"answer":[115],"this,":[116],"design":[118],"two":[119],"operators:":[120],"TQuant":[121],"MQuant":[123],"correspond":[125],"these":[127],"respective":[128],"minimization":[129],"tasks.":[130],"We":[131],"show":[132],"experimentally":[133],"our":[135],"approach":[136],"allows":[137],"significantly":[139],"improve":[140],"through":[146],"variety":[148],"scenarios":[150],"DFQ,":[152],"PTQ":[153],"QAT":[155],"give":[157],"strong":[158],"insights":[159],"pave":[161],"way":[163],"future":[165],"research":[166],"deep":[168],"network":[170],"quantization.":[171]},"counts_by_year":[],"updated_date":"2026-06-22T08:00:12.763002","created_date":"2025-10-10T00:00:00"}