{"id":"https://openalex.org/W3109606446","doi":"https://doi.org/10.1142/s0217595923400055","title":"Eigenvalue-Corrected Natural Gradient Based on a New Approximation","display_name":"Eigenvalue-Corrected Natural Gradient Based on a New Approximation","publication_year":2023,"publication_date":"2023-01-13","ids":{"openalex":"https://openalex.org/W3109606446","doi":"https://doi.org/10.1142/s0217595923400055","mag":"3109606446"},"language":"en","primary_location":{"id":"doi:10.1142/s0217595923400055","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1142/s0217595923400055","pdf_url":null,"source":{"id":"https://openalex.org/S1254427","display_name":"Asia Pacific Journal of Operational Research","issn_l":"0217-5959","issn":["0217-5959","1793-7019"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319815","host_organization_name":"World Scientific","host_organization_lineage":["https://openalex.org/P4310319815"],"host_organization_lineage_names":["World Scientific"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Asia-Pacific Journal of Operational Research","raw_type":"journal-article"},"type":"article","indexed_in":["arxiv","crossref","datacite"],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://arxiv.org/pdf/2011.13609","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5101650297","display_name":"Kaixin Gao","orcid":"https://orcid.org/0000-0003-3728-2294"},"institutions":[{"id":"https://openalex.org/I162868743","display_name":"Tianjin University","ror":"https://ror.org/012tb2g32","country_code":"CN","type":"education","lineage":["https://openalex.org/I162868743"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Kaixin Gao","raw_affiliation_strings":["School of Mathematics, Tianjin University, Tianjin 300350, P. R. China"],"affiliations":[{"raw_affiliation_string":"School of Mathematics, Tianjin University, Tianjin 300350, P. R. China","institution_ids":["https://openalex.org/I162868743"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5050088762","display_name":"Zheng\u2010Hai Huang","orcid":"https://orcid.org/0000-0003-2269-961X"},"institutions":[{"id":"https://openalex.org/I162868743","display_name":"Tianjin University","ror":"https://ror.org/012tb2g32","country_code":"CN","type":"education","lineage":["https://openalex.org/I162868743"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zheng-Hai Huang","raw_affiliation_strings":["School of Mathematics, Tianjin University, Tianjin 300350, P. R. China"],"affiliations":[{"raw_affiliation_string":"School of Mathematics, Tianjin University, Tianjin 300350, P. R. China","institution_ids":["https://openalex.org/I162868743"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100442739","display_name":"Xiaolei Liu","orcid":"https://orcid.org/0000-0001-8510-4025"},"institutions":[{"id":"https://openalex.org/I162868743","display_name":"Tianjin University","ror":"https://ror.org/012tb2g32","country_code":"CN","type":"education","lineage":["https://openalex.org/I162868743"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Xiaolei Liu","raw_affiliation_strings":["School of Mathematics, Tianjin University, Tianjin 300350, P. R. China"],"affiliations":[{"raw_affiliation_string":"School of Mathematics, Tianjin University, Tianjin 300350, P. R. China","institution_ids":["https://openalex.org/I162868743"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100340937","display_name":"Min Wang","orcid":"https://orcid.org/0000-0002-7676-2285"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Min Wang","raw_affiliation_strings":["Central Software Institute, Huawei Technologies Co. Ltd, Hangzhou 310051, P. R. China"],"affiliations":[{"raw_affiliation_string":"Central Software Institute, Huawei Technologies Co. Ltd, Hangzhou 310051, P. R. China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5016454888","display_name":"Shuangling Wang","orcid":"https://orcid.org/0000-0002-4243-3555"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Shuangling Wang","raw_affiliation_strings":["Central Software Institute, Huawei Technologies Co. Ltd, Hangzhou 310051, P. R. China"],"affiliations":[{"raw_affiliation_string":"Central Software Institute, Huawei Technologies Co. Ltd, Hangzhou 310051, P. R. China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100699858","display_name":"Zidong Wang","orcid":"https://orcid.org/0000-0002-9576-7401"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Zidong Wang","raw_affiliation_strings":["Central Software Institute, Huawei Technologies Co. Ltd, Hangzhou 310051, P. R. China"],"affiliations":[{"raw_affiliation_string":"Central Software Institute, Huawei Technologies Co. Ltd, Hangzhou 310051, P. R. China","institution_ids":["https://openalex.org/I2250955327"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5100341936","display_name":"Dachuan Xu","orcid":"https://orcid.org/0000-0002-7846-0969"},"institutions":[{"id":"https://openalex.org/I37796252","display_name":"Beijing University of Technology","ror":"https://ror.org/037b1pp87","country_code":"CN","type":"education","lineage":["https://openalex.org/I37796252"]}],"countries":["CN"],"is_corresponding":true,"raw_author_name":"Dachuan Xu","raw_affiliation_strings":["Beijing Institute for Scientific and Engineering Computing, Beijing University of Technology, Beijing 100124, P. R. China"],"affiliations":[{"raw_affiliation_string":"Beijing Institute for Scientific and Engineering Computing, Beijing University of Technology, Beijing 100124, P. R. China","institution_ids":["https://openalex.org/I37796252"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5100616400","display_name":"Fan Yu","orcid":"https://orcid.org/0000-0002-0576-7396"},"institutions":[{"id":"https://openalex.org/I2250955327","display_name":"Huawei Technologies (China)","ror":"https://ror.org/00cmhce21","country_code":"CN","type":"company","lineage":["https://openalex.org/I2250955327"]}],"countries":["CN"],"is_corresponding":false,"raw_author_name":"Fan Yu","raw_affiliation_strings":["Central Software Institute, Huawei Technologies Co. Ltd, Hangzhou 310051, P. R. China"],"affiliations":[{"raw_affiliation_string":"Central Software Institute, Huawei Technologies Co. Ltd, Hangzhou 310051, P. R. China","institution_ids":["https://openalex.org/I2250955327"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":8,"corresponding_author_ids":["https://openalex.org/A5100341936"],"corresponding_institution_ids":["https://openalex.org/I37796252"],"apc_list":null,"apc_paid":null,"fwci":0.0,"has_fulltext":false,"cited_by_count":1,"citation_normalized_percentile":{"value":0.00169014,"is_in_top_1_percent":false,"is_in_top_10_percent":false},"cited_by_percentile_year":{"min":89,"max":94},"biblio":{"volume":"40","issue":"01","first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T11447","display_name":"Blind Source Separation Techniques","score":0.9998000264167786,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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/T11447","display_name":"Blind Source Separation Techniques","score":0.9998000264167786,"subfield":{"id":"https://openalex.org/subfields/1711","display_name":"Signal Processing"},"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":0.9983000159263611,"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/T10057","display_name":"Face and Expression Recognition","score":0.991100013256073,"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/kronecker-delta","display_name":"Kronecker delta","score":0.7732641696929932},{"id":"https://openalex.org/keywords/trace","display_name":"TRACE (psycholinguistics)","score":0.663200855255127},{"id":"https://openalex.org/keywords/scaling","display_name":"Scaling","score":0.5885963439941406},{"id":"https://openalex.org/keywords/factorization","display_name":"Factorization","score":0.5669330954551697},{"id":"https://openalex.org/keywords/diagonal","display_name":"Diagonal","score":0.5293002128601074},{"id":"https://openalex.org/keywords/mathematics","display_name":"Mathematics","score":0.5285785794258118},{"id":"https://openalex.org/keywords/applied-mathematics","display_name":"Applied mathematics","score":0.5255614519119263},{"id":"https://openalex.org/keywords/kronecker-product","display_name":"Kronecker product","score":0.5148500800132751},{"id":"https://openalex.org/keywords/eigenvalues-and-eigenvectors","display_name":"Eigenvalues and eigenvectors","score":0.5143734216690063},{"id":"https://openalex.org/keywords/matrix-decomposition","display_name":"Matrix decomposition","score":0.4290889501571655},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.2793292701244354},{"id":"https://openalex.org/keywords/geometry","display_name":"Geometry","score":0.08433270454406738},{"id":"https://openalex.org/keywords/physics","display_name":"Physics","score":0.08382165431976318}],"concepts":[{"id":"https://openalex.org/C39482219","wikidata":"https://www.wikidata.org/wiki/Q192826","display_name":"Kronecker delta","level":2,"score":0.7732641696929932},{"id":"https://openalex.org/C75291252","wikidata":"https://www.wikidata.org/wiki/Q1315756","display_name":"TRACE (psycholinguistics)","level":2,"score":0.663200855255127},{"id":"https://openalex.org/C99844830","wikidata":"https://www.wikidata.org/wiki/Q102441924","display_name":"Scaling","level":2,"score":0.5885963439941406},{"id":"https://openalex.org/C187834632","wikidata":"https://www.wikidata.org/wiki/Q188804","display_name":"Factorization","level":2,"score":0.5669330954551697},{"id":"https://openalex.org/C130367717","wikidata":"https://www.wikidata.org/wiki/Q189791","display_name":"Diagonal","level":2,"score":0.5293002128601074},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.5285785794258118},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.5255614519119263},{"id":"https://openalex.org/C46030957","wikidata":"https://www.wikidata.org/wiki/Q1238125","display_name":"Kronecker product","level":3,"score":0.5148500800132751},{"id":"https://openalex.org/C158693339","wikidata":"https://www.wikidata.org/wiki/Q190524","display_name":"Eigenvalues and eigenvectors","level":2,"score":0.5143734216690063},{"id":"https://openalex.org/C42355184","wikidata":"https://www.wikidata.org/wiki/Q1361088","display_name":"Matrix decomposition","level":3,"score":0.4290889501571655},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.2793292701244354},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.08433270454406738},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.08382165431976318},{"id":"https://openalex.org/C41895202","wikidata":"https://www.wikidata.org/wiki/Q8162","display_name":"Linguistics","level":1,"score":0.0},{"id":"https://openalex.org/C138885662","wikidata":"https://www.wikidata.org/wiki/Q5891","display_name":"Philosophy","level":0,"score":0.0},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0}],"mesh":[],"locations_count":4,"locations":[{"id":"doi:10.1142/s0217595923400055","is_oa":false,"landing_page_url":"http://dx.doi.org/10.1142/s0217595923400055","pdf_url":null,"source":{"id":"https://openalex.org/S1254427","display_name":"Asia Pacific Journal of Operational Research","issn_l":"0217-5959","issn":["0217-5959","1793-7019"],"is_oa":false,"is_in_doaj":false,"is_core":true,"host_organization":"https://openalex.org/P4310319815","host_organization_name":"World Scientific","host_organization_lineage":["https://openalex.org/P4310319815"],"host_organization_lineage_names":["World Scientific"],"type":"journal"},"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"Asia-Pacific Journal of Operational Research","raw_type":"journal-article"},{"id":"pmh:oai:arXiv.org:2011.13609","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2011.13609","pdf_url":"https://arxiv.org/pdf/2011.13609","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},{"id":"pmh:oai:RePEc:wsi:apjorx:v:40:y:2023:i:01:n:s0217595923400055","is_oa":false,"landing_page_url":"http://www.worldscientific.com/doi/abs/10.1142/S0217595923400055","pdf_url":null,"source":{"id":"https://openalex.org/S4306401271","display_name":"RePEc: Research Papers in Economics","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I77793887","host_organization_name":"Federal Reserve Bank of St. Louis","host_organization_lineage":["https://openalex.org/I77793887"],"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":"article"},{"id":"doi:10.48550/arxiv.2011.13609","is_oa":true,"landing_page_url":"https://doi.org/10.48550/arxiv.2011.13609","pdf_url":null,"source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":null,"raw_source_name":null,"raw_type":"article"}],"best_oa_location":{"id":"pmh:oai:arXiv.org:2011.13609","is_oa":true,"landing_page_url":"http://arxiv.org/abs/2011.13609","pdf_url":"https://arxiv.org/pdf/2011.13609","source":{"id":"https://openalex.org/S4306400194","display_name":"arXiv (Cornell University)","issn_l":null,"issn":null,"is_oa":true,"is_in_doaj":false,"is_core":false,"host_organization":"https://openalex.org/I205783295","host_organization_name":"Cornell University","host_organization_lineage":["https://openalex.org/I205783295"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"text"},"sustainable_development_goals":[],"awards":[{"id":"https://openalex.org/G2099803721","display_name":null,"funder_award_id":"11871051","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":29,"referenced_works":["https://openalex.org/W196761320","https://openalex.org/W1686810756","https://openalex.org/W1970789124","https://openalex.org/W1980287119","https://openalex.org/W1994616650","https://openalex.org/W2062294426","https://openalex.org/W2102017903","https://openalex.org/W2146502635","https://openalex.org/W2194775991","https://openalex.org/W2335728318","https://openalex.org/W2580041566","https://openalex.org/W2750933313","https://openalex.org/W2786263712","https://openalex.org/W2902363477","https://openalex.org/W2913310393","https://openalex.org/W2962739609","https://openalex.org/W2963241285","https://openalex.org/W2963309496","https://openalex.org/W2963344792","https://openalex.org/W2963746765","https://openalex.org/W2964121744","https://openalex.org/W2964125128","https://openalex.org/W2967791890","https://openalex.org/W2990138404","https://openalex.org/W3035182906","https://openalex.org/W3035343021","https://openalex.org/W3039833373","https://openalex.org/W3118608800","https://openalex.org/W3173304480"],"related_works":["https://openalex.org/W123417539","https://openalex.org/W4233239985","https://openalex.org/W2079407403","https://openalex.org/W2060299328","https://openalex.org/W4375957393","https://openalex.org/W2117336295","https://openalex.org/W2073634779","https://openalex.org/W3102722572","https://openalex.org/W2090103374","https://openalex.org/W1564288589"],"abstract_inverted_index":{"Using":[0],"second-order":[1],"optimization":[2],"methods":[3,105],"for":[4],"training":[5],"deep":[6],"neural":[7],"networks":[8],"(DNNs)":[9],"has":[10],"attracted":[11],"many":[12],"researchers.":[13],"A":[14],"recently":[15],"proposed":[16,22,114],"method,":[17],"Eigenvalue-corrected":[18,109],"Kronecker":[19,78,110],"Factorization":[20,111],"(EKFAC),":[21],"an":[23],"interpretation":[24],"by":[25,76,140],"viewing":[26],"natural":[27,52],"gradient":[28,53],"update":[29],"as":[30,72],"a":[31,46,73],"diagonal":[32],"method":[33,48,115,133,156],"and":[34,83,91,106,134,147,163],"corrects":[35,118],"the":[36,41,51,65,77,84,93,100,119,124,130,135,145,150],"inaccurate":[37],"re-scaling":[38,121],"factor":[39,122],"in":[40,63],"KFAC":[42],"eigenbasis.":[43],"What\u2019s":[44],"more,":[45],"new":[47,131],"to":[49],"approximate":[50],"called":[54],"Trace-restricted":[55,108],"Kronecker-factored":[56,125,152],"Approximate":[57],"Curvature":[58],"(TKFAC)":[59],"is":[60,70],"also":[61,128,143],"proposed,":[62],"which":[64],"Fisher":[66],"information":[67],"matrix":[68],"(FIM)":[69],"approximated":[71],"constant":[74],"multiplied":[75],"product":[79],"of":[80,102],"two":[81,104],"matrices":[82],"traces":[85],"can":[86],"be":[87],"kept":[88],"equal":[89],"before":[90],"after":[92],"approximation.":[94],"In":[95],"this":[96],"work,":[97],"we":[98],"combine":[99],"ideas":[101],"these":[103],"propose":[107],"(TEKFAC).":[112],"The":[113],"not":[116],"only":[117],"inexact":[120],"under":[123],"eigenbasis,":[126],"but":[127],"considers":[129],"approximation":[132],"effective":[136],"damping":[137],"technique":[138],"adopted":[139],"TKFAC.":[141],"We":[142],"discuss":[144],"differences":[146],"relationships":[148],"among":[149],"related":[151],"approximations.":[153],"Empirically,":[154],"our":[155],"outperforms":[157],"SGD":[158],"with":[159],"momentum,":[160],"Adam,":[161],"EKFAC":[162],"TKFAC":[164],"on":[165],"several":[166],"DNNs.":[167]},"counts_by_year":[{"year":2020,"cited_by_count":1}],"updated_date":"2025-12-23T23:11:35.936235","created_date":"2020-12-07T00:00:00"}