{"id":"https://openalex.org/W2965380306","doi":"https://doi.org/10.24963/ijcai.2019/425","title":"Learning Low-precision Neural Networks without Straight-Through Estimator (STE)","display_name":"Learning Low-precision Neural Networks without Straight-Through Estimator (STE)","publication_year":2019,"publication_date":"2019-07-28","ids":{"openalex":"https://openalex.org/W2965380306","doi":"https://doi.org/10.24963/ijcai.2019/425","mag":"2965380306"},"language":"en","primary_location":{"id":"doi:10.24963/ijcai.2019/425","is_oa":true,"landing_page_url":"https://doi.org/10.24963/ijcai.2019/425","pdf_url":"https://www.ijcai.org/proceedings/2019/0425.pdf","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence","raw_type":"proceedings-article"},"type":"article","indexed_in":["crossref"],"open_access":{"is_oa":true,"oa_status":"gold","oa_url":"https://www.ijcai.org/proceedings/2019/0425.pdf","any_repository_has_fulltext":null},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5112668548","display_name":"Zhi-Gang Liu","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":true,"raw_author_name":"Zhi-Gang Liu","raw_affiliation_strings":["Arm Machine Learning Research Lab"],"affiliations":[{"raw_affiliation_string":"Arm Machine Learning Research Lab","institution_ids":["https://openalex.org/I4210156213"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5090422974","display_name":"Matthew Mattina","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":"Matthew Mattina","raw_affiliation_strings":["Arm Machine Learning Research Lab"],"affiliations":[{"raw_affiliation_string":"Arm Machine Learning Research Lab","institution_ids":["https://openalex.org/I4210156213"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":2,"corresponding_author_ids":["https://openalex.org/A5112668548"],"corresponding_institution_ids":["https://openalex.org/I4210156213"],"apc_list":null,"apc_paid":null,"fwci":2.733,"has_fulltext":false,"cited_by_count":37,"citation_normalized_percentile":{"value":0.92526037,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":94,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"3066","last_page":"3072"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10036","display_name":"Advanced Neural Network Applications","score":0.9994999766349792,"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":0.9994999766349792,"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":0.9991000294685364,"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/T11609","display_name":"Geophysical Methods and Applications","score":0.9979000091552734,"subfield":{"id":"https://openalex.org/subfields/2212","display_name":"Ocean Engineering"},"field":{"id":"https://openalex.org/fields/22","display_name":"Engineering"},"domain":{"id":"https://openalex.org/domains/3","display_name":"Physical Sciences"}}],"keywords":[{"id":"https://openalex.org/keywords/estimator","display_name":"Estimator","score":0.69420325756073},{"id":"https://openalex.org/keywords/stochastic-gradient-descent","display_name":"Stochastic gradient descent","score":0.6703469753265381},{"id":"https://openalex.org/keywords/artificial-neural-network","display_name":"Artificial neural network","score":0.6398431062698364},{"id":"https://openalex.org/keywords/quantization","display_name":"Quantization (signal processing)","score":0.6267139911651611},{"id":"https://openalex.org/keywords/affine-transformation","display_name":"Affine transformation","score":0.6130492091178894},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.5944325923919678},{"id":"https://openalex.org/keywords/gradient-descent","display_name":"Gradient descent","score":0.5462935566902161},{"id":"https://openalex.org/keywords/affine-combination","display_name":"Affine combination","score":0.48502036929130554},{"id":"https://openalex.org/keywords/residual-neural-network","display_name":"Residual neural network","score":0.4823884069919586},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.45515841245651245},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.42201071977615356},{"id":"https://openalex.org/keywords/artificial-intelligence","display_name":"Artificial intelligence","score":0.36571723222732544},{"id":"https://openalex.org/keywords/pattern-recognition","display_name":"Pattern recognition (psychology)","score":0.3288063704967499},{"id":"https://openalex.org/keywords/statistics","display_name":"Statistics","score":0.18275931477546692},{"id":"https://openalex.org/keywords/geometry","display_name":"Geometry","score":0.0923440158367157}],"concepts":[{"id":"https://openalex.org/C185429906","wikidata":"https://www.wikidata.org/wiki/Q1130160","display_name":"Estimator","level":2,"score":0.69420325756073},{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.6703469753265381},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.6398431062698364},{"id":"https://openalex.org/C28855332","wikidata":"https://www.wikidata.org/wiki/Q198099","display_name":"Quantization (signal processing)","level":2,"score":0.6267139911651611},{"id":"https://openalex.org/C92757383","wikidata":"https://www.wikidata.org/wiki/Q382497","display_name":"Affine transformation","level":2,"score":0.6130492091178894},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.5944325923919678},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.5462935566902161},{"id":"https://openalex.org/C207221997","wikidata":"https://www.wikidata.org/wiki/Q938614","display_name":"Affine combination","level":3,"score":0.48502036929130554},{"id":"https://openalex.org/C2944601119","wikidata":"https://www.wikidata.org/wiki/Q43744058","display_name":"Residual neural network","level":3,"score":0.4823884069919586},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.45515841245651245},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.42201071977615356},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.36571723222732544},{"id":"https://openalex.org/C153180895","wikidata":"https://www.wikidata.org/wiki/Q7148389","display_name":"Pattern recognition (psychology)","level":2,"score":0.3288063704967499},{"id":"https://openalex.org/C105795698","wikidata":"https://www.wikidata.org/wiki/Q12483","display_name":"Statistics","level":1,"score":0.18275931477546692},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.0923440158367157}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.24963/ijcai.2019/425","is_oa":true,"landing_page_url":"https://doi.org/10.24963/ijcai.2019/425","pdf_url":"https://www.ijcai.org/proceedings/2019/0425.pdf","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence","raw_type":"proceedings-article"}],"best_oa_location":{"id":"doi:10.24963/ijcai.2019/425","is_oa":true,"landing_page_url":"https://doi.org/10.24963/ijcai.2019/425","pdf_url":"https://www.ijcai.org/proceedings/2019/0425.pdf","source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence","raw_type":"proceedings-article"},"sustainable_development_goals":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W2965380306.pdf","grobid_xml":"https://content.openalex.org/works/W2965380306.grobid-xml"},"referenced_works_count":17,"referenced_works":["https://openalex.org/W1902934009","https://openalex.org/W2160815625","https://openalex.org/W2242818861","https://openalex.org/W2279098554","https://openalex.org/W2294543795","https://openalex.org/W2300242332","https://openalex.org/W2613332842","https://openalex.org/W2618530766","https://openalex.org/W2739789140","https://openalex.org/W2809624076","https://openalex.org/W2898827265","https://openalex.org/W2963082137","https://openalex.org/W2964123455","https://openalex.org/W2964228333","https://openalex.org/W4295262505","https://openalex.org/W4297775537","https://openalex.org/W4375869345"],"related_works":["https://openalex.org/W2038416447","https://openalex.org/W2073801174","https://openalex.org/W2354159744","https://openalex.org/W2077614002","https://openalex.org/W2100544925","https://openalex.org/W4206903459","https://openalex.org/W1965628666","https://openalex.org/W2754816816","https://openalex.org/W2773810497","https://openalex.org/W4366280654"],"abstract_inverted_index":{"The":[0],"Straight-Through":[1],"Estimator":[2],"(STE)":[3],"is":[4,84,112],"widely":[5],"used":[6],"for":[7],"back-propagating":[8],"gradients":[9],"through":[10,98],"the":[11,15,51,55,63,68,91,95,99,107,110,122,143,147,163],"quantization":[12],"function,":[13],"but":[14],"STE":[16,47,166],"technique":[17],"lacks":[18],"a":[19,125],"complete":[20],"theoretical":[21],"understanding.":[22],"We":[23],"propose":[24],"an":[25,59],"alternative":[26],"methodology":[27],"called":[28],"alpha-blending":[29,144],"(AB),":[30],"which":[31],"quantizes":[32],"neural":[33],"networks":[34],"to":[35,89,94,116,162],"low":[36,117],"precision":[37,101,118],"using":[38,142],"stochastic":[39],"gradient":[40,92],"descent":[41],"(SGD).":[42],"Our":[43],"AB":[44,123,151],"method":[45],"avoids":[46],"approximation":[48],"by":[49,58,155],"replacing":[50],"quantized":[52,64],"weight":[53,65,71],"in":[54],"loss":[56],"function":[57],"affine":[60,108],"combination":[61],"of":[62,106,165],"w_q":[66],"and":[67,78,131,146,158],"corresponding":[69],"full-precision":[70,115],"w":[72],"with":[73],"non-trainable":[74],"scalar":[75],"coefficient":[76],"alpha":[77,83],"(1-":[79],"alpha).":[80],"During":[81],"training,":[82],"gradually":[85],"increased":[86],"from":[87,114],"0":[88],"1;":[90],"updates":[93],"weights":[96],"are":[97,140],"full":[100],"term,":[102],"(1-alpha)":[103],"*":[104],"w,":[105],"combination;":[109],"model":[111],"converted":[113],"progressively.":[119],"To":[120],"evaluate":[121],"method,":[124],"1-bit":[126],"BinaryNet":[127],"on":[128,138],"CIFAR10":[129],"dataset":[130],"8-bits,":[132],"4-bits":[133],"MobileNet":[134],"v1,":[135],"ResNet_50":[136],"v1/2":[137],"ImageNet":[139],"trained":[141],"approach,":[145],"evaluation":[148],"indicates":[149],"that":[150],"improves":[152],"top-1":[153],"accuracy":[154],"0.9\\%,":[156],"0.82\\%":[157],"2.93\\%":[159],"respectively":[160],"compared":[161],"results":[164],"based":[167],"quantization.":[168]},"counts_by_year":[{"year":2025,"cited_by_count":3},{"year":2024,"cited_by_count":5},{"year":2023,"cited_by_count":2},{"year":2022,"cited_by_count":7},{"year":2021,"cited_by_count":8},{"year":2020,"cited_by_count":10},{"year":2019,"cited_by_count":2}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}