{"id":"https://openalex.org/W2736068458","doi":"https://doi.org/10.1109/ijcnn.2017.7966324","title":"On-chip training of recurrent neural networks with limited numerical precision","display_name":"On-chip training of recurrent neural networks with limited numerical precision","publication_year":2017,"publication_date":"2017-05-01","ids":{"openalex":"https://openalex.org/W2736068458","doi":"https://doi.org/10.1109/ijcnn.2017.7966324","mag":"2736068458"},"language":"en","primary_location":{"id":"doi:10.1109/ijcnn.2017.7966324","is_oa":false,"landing_page_url":"https://doi.org/10.1109/ijcnn.2017.7966324","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2017 International Joint Conference on Neural Networks (IJCNN)","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/A5002187268","display_name":"Taesik Na","orcid":"https://orcid.org/0000-0003-2039-3772"},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":true,"raw_author_name":"Taesik Na","raw_affiliation_strings":["School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA"],"affiliations":[{"raw_affiliation_string":"School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5063908850","display_name":"Jong Hwan Ko","orcid":"https://orcid.org/0000-0003-4434-4318"},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jong Hwan Ko","raw_affiliation_strings":["School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA"],"affiliations":[{"raw_affiliation_string":"School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"middle","author":{"id":"https://openalex.org/A5079872127","display_name":"Jaeha Kung","orcid":"https://orcid.org/0000-0001-6151-8602"},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Jaeha Kung","raw_affiliation_strings":["School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA"],"affiliations":[{"raw_affiliation_string":"School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA","institution_ids":["https://openalex.org/I130701444"]}]},{"author_position":"last","author":{"id":"https://openalex.org/A5009591041","display_name":"Saibal Mukhopadhyay","orcid":"https://orcid.org/0000-0002-8894-3390"},"institutions":[{"id":"https://openalex.org/I130701444","display_name":"Georgia Institute of Technology","ror":"https://ror.org/01zkghx44","country_code":"US","type":"education","lineage":["https://openalex.org/I130701444"]}],"countries":["US"],"is_corresponding":false,"raw_author_name":"Saibal Mukhopadhyay","raw_affiliation_strings":["School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA"],"affiliations":[{"raw_affiliation_string":"School of ECE, Georgia Institute of Technology, Atlanta, Georgia, USA","institution_ids":["https://openalex.org/I130701444"]}]}],"institutions":[],"countries_distinct_count":1,"institutions_distinct_count":4,"corresponding_author_ids":["https://openalex.org/A5002187268"],"corresponding_institution_ids":["https://openalex.org/I130701444"],"apc_list":null,"apc_paid":null,"fwci":2.7303,"has_fulltext":false,"cited_by_count":42,"citation_normalized_percentile":{"value":0.92404349,"is_in_top_1_percent":false,"is_in_top_10_percent":true},"cited_by_percentile_year":{"min":90,"max":99},"biblio":{"volume":null,"issue":null,"first_page":"3716","last_page":"3723"},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10320","display_name":"Neural Networks and Applications","score":0.9993000030517578,"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/T10320","display_name":"Neural Networks and Applications","score":0.9993000030517578,"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/T12808","display_name":"Ferroelectric and Negative Capacitance Devices","score":0.9986000061035156,"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/T10036","display_name":"Advanced Neural Network Applications","score":0.9983000159263611,"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/rounding","display_name":"Rounding","score":0.8701488375663757},{"id":"https://openalex.org/keywords/computer-science","display_name":"Computer science","score":0.8009469509124756},{"id":"https://openalex.org/keywords/floating-point","display_name":"Floating point","score":0.5785072445869446},{"id":"https://openalex.org/keywords/algorithm","display_name":"Algorithm","score":0.48677706718444824},{"id":"https://openalex.org/keywords/shift-register","display_name":"Shift register","score":0.4569322466850281},{"id":"https://openalex.org/keywords/computer-hardware","display_name":"Computer hardware","score":0.45216241478919983},{"id":"https://openalex.org/keywords/chip","display_name":"Chip","score":0.2813010811805725}],"concepts":[{"id":"https://openalex.org/C136625980","wikidata":"https://www.wikidata.org/wiki/Q663208","display_name":"Rounding","level":2,"score":0.8701488375663757},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.8009469509124756},{"id":"https://openalex.org/C84211073","wikidata":"https://www.wikidata.org/wiki/Q117879","display_name":"Floating point","level":2,"score":0.5785072445869446},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.48677706718444824},{"id":"https://openalex.org/C49654631","wikidata":"https://www.wikidata.org/wiki/Q746165","display_name":"Shift register","level":3,"score":0.4569322466850281},{"id":"https://openalex.org/C9390403","wikidata":"https://www.wikidata.org/wiki/Q3966","display_name":"Computer hardware","level":1,"score":0.45216241478919983},{"id":"https://openalex.org/C165005293","wikidata":"https://www.wikidata.org/wiki/Q1074500","display_name":"Chip","level":2,"score":0.2813010811805725},{"id":"https://openalex.org/C76155785","wikidata":"https://www.wikidata.org/wiki/Q418","display_name":"Telecommunications","level":1,"score":0.0},{"id":"https://openalex.org/C111919701","wikidata":"https://www.wikidata.org/wiki/Q9135","display_name":"Operating system","level":1,"score":0.0}],"mesh":[],"locations_count":1,"locations":[{"id":"doi:10.1109/ijcnn.2017.7966324","is_oa":false,"landing_page_url":"https://doi.org/10.1109/ijcnn.2017.7966324","pdf_url":null,"source":null,"license":null,"license_id":null,"version":"publishedVersion","is_accepted":true,"is_published":true,"raw_source_name":"2017 International Joint Conference on Neural Networks (IJCNN)","raw_type":"proceedings-article"}],"best_oa_location":null,"sustainable_development_goals":[{"id":"https://metadata.un.org/sdg/7","score":0.8199999928474426,"display_name":"Affordable and clean energy"}],"awards":[],"funders":[{"id":"https://openalex.org/F4320306076","display_name":"National Science Foundation","ror":"https://ror.org/021nxhr62"},{"id":"https://openalex.org/F4320337345","display_name":"Office of Naval Research","ror":"https://ror.org/00rk2pe57"}],"has_content":{"grobid_xml":false,"pdf":false},"content_urls":null,"referenced_works_count":24,"referenced_works":["https://openalex.org/W1815076433","https://openalex.org/W1836465849","https://openalex.org/W1855892484","https://openalex.org/W2017369466","https://openalex.org/W2064675550","https://openalex.org/W2106033855","https://openalex.org/W2119799051","https://openalex.org/W2130942839","https://openalex.org/W2145339207","https://openalex.org/W2193413348","https://openalex.org/W2251682575","https://openalex.org/W2254715784","https://openalex.org/W2257979135","https://openalex.org/W2475884125","https://openalex.org/W2556936974","https://openalex.org/W2962949994","https://openalex.org/W2963251942","https://openalex.org/W2964199361","https://openalex.org/W3196022191","https://openalex.org/W4253802691","https://openalex.org/W4300769538","https://openalex.org/W4302375066","https://openalex.org/W6679436768","https://openalex.org/W6691839725"],"related_works":["https://openalex.org/W2535419677","https://openalex.org/W2134876043","https://openalex.org/W2476214559","https://openalex.org/W3039927303","https://openalex.org/W4388911419","https://openalex.org/W4322008031","https://openalex.org/W2544512391","https://openalex.org/W2118569990","https://openalex.org/W2389154333","https://openalex.org/W1989068301"],"abstract_inverted_index":{"Training":[0],"of":[1,27,117,193],"neural":[2],"network":[3],"can":[4,22,60,142],"be":[5,81],"accelerated":[6],"by":[7,125],"limited":[8],"numerical":[9],"precision":[10,21,32,67,91,120,129,152,177],"together":[11],"with":[12,65,93,139,147,157,164],"specialized":[13],"low-precision":[14],"hardware.":[15,196],"This":[16],"paper":[17],"studies":[18],"how":[19],"low":[20,31,66,90,119,128,176],"impact":[23],"on":[24,57],"entire":[25],"training":[26,33,64,92,121,145],"RNNs.":[28],"We":[29,51,73,88,131],"emulate":[30],"for":[34,84,168],"recently":[35],"proposed":[36],"gated":[37],"recurrent":[38],"unit":[39],"(GRU)":[40],"and":[41,103,154,170,182],"use":[42],"dynamic":[43,85],"fixed":[44,86],"point":[45,149,195],"as":[46,68,70],"a":[47],"target":[48],"numeric":[49],"format.":[50],"first":[52],"show":[53,75,132,175],"that":[54,76,133,192],"batch":[55],"normalization":[56],"input":[58],"sequences":[59],"help":[61],"speed":[62],"up":[63,187],"well":[69],"high":[71],"precision.":[72,150],"also":[74,123],"the":[77,112,134],"overflow":[78],"rate":[79],"should":[80],"carefully":[82],"controlled":[83],"point.":[87],"study":[89],"various":[94],"rounding":[95,107,141],"options":[96],"including":[97],"bit":[98],"truncation,":[99],"round":[100],"to":[101,188],"nearest,":[102],"stochastic":[104,140],"rounding.":[105],"Stochastic":[106],"shows":[108],"superior":[109],"results":[110,146,174],"than":[111,191],"other":[113],"options.":[114],"The":[115],"effect":[116],"fully":[118],"is":[122,162,179,186],"analyzed":[124],"comparing":[126],"partial":[127],"training.":[130],"piecewise":[135],"linear":[136],"activation":[137],"function":[138],"achieve":[143],"comparable":[144],"floating":[148,194],"Low":[151],"multiplier":[153],"accumulator":[155],"(MAC)":[156],"linear-feedback":[158],"shift":[159],"register":[160],"(LFSR)":[161],"implemented":[163],"28nm":[165],"Synopsys":[166],"PDK":[167],"energy":[169,183],"performance":[171],"analysis.":[172],"Implementation":[173],"hardware":[178],"4.7\u00d7":[180],"faster,":[181],"per":[184],"task":[185],"4.55\u00d7":[189],"lower":[190]},"counts_by_year":[{"year":2025,"cited_by_count":6},{"year":2024,"cited_by_count":1},{"year":2023,"cited_by_count":6},{"year":2022,"cited_by_count":8},{"year":2021,"cited_by_count":7},{"year":2020,"cited_by_count":4},{"year":2019,"cited_by_count":6},{"year":2018,"cited_by_count":3},{"year":2017,"cited_by_count":1}],"updated_date":"2025-11-06T03:46:38.306776","created_date":"2025-10-10T00:00:00"}