{"id":"https://openalex.org/W2963992014","doi":null,"title":"Efficient orthogonal parametrisation of recurrent neural networks using householder reflections","display_name":"Efficient orthogonal parametrisation of recurrent neural networks using householder reflections","alternate_titles":[],"publication_year":2017,"publication_date":"2017-07-17","ids":{"openalex":"https://openalex.org/W2963992014","mag":"2963992014"},"language":"en","primary_location":{"id":"pmh:oai:jupiter.its.unimelb.edu.au:11343/192906","doi":null,"is_oa":true,"landing_page_url":"http://hdl.handle.net/11343/192906","pdf_url":null,"source":{"id":"https://openalex.org/S4377196259","display_name":"Minerva Access (University of Melbourne)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I165779595","host_organization_name":"The University of Melbourne","host_organization_lineage":["https://openalex.org/I165779595"],"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":"Masters Research thesis"},"parent_works":[],"lineage":["https://openalex.org/W2963992014"],"child_works_count":0,"type":"dissertation","is_traditional":true,"indexed_in":[],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"http://hdl.handle.net/11343/192906","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5110662310","display_name":"Zakaria Mhammedi","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Zakaria Mhammedi","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5033538273","display_name":"Andrew Hellicar","orcid":null,"observed_orcids":[]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Andrew Hellicar","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]},{"author_position":"middle","author":{"id":"https://openalex.org/A5000151716","display_name":"Ashfaqur Rahman","orcid":"https://orcid.org/0000-0002-0236-5447","observed_orcids":["https://orcid.org/0000-0002-0236-5447"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"Ashfaqur Rahman","raw_affiliation_strings":[],"raw_orcid":"https://orcid.org/0000-0002-0236-5447","affiliations":[]},{"author_position":"last","author":{"id":"https://openalex.org/A5101609697","display_name":"James Bailey","orcid":"https://orcid.org/0000-0002-3769-3811","observed_orcids":["https://orcid.org/0000-0002-3769-3811"]},"institutions":[],"countries":[],"is_corresponding":false,"raw_author_name":"James Bailey","raw_affiliation_strings":[],"raw_orcid":"https://orcid.org/0000-0002-3769-3811","affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":[],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":139,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":"70","issue":null,"first_page":"2401","last_page":"2409"},"is_retracted":false,"is_pulled":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10320","display_name":"Neural Networks and Applications","score":0.9672999978065491,"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":0.9672999978065491,"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/T10028","display_name":"Topic Modeling","score":0.0026000000070780516,"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/T10792","display_name":"Matrix Theory and Algorithms","score":0.002400000113993883,"subfield":{"id":"https://openalex.org/subfields/1703","display_name":"Computational Theory and Mathematics"},"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/recurrent-neural-networks","display_name":"recurrent neural networks","score":0.9919999837875366},{"id":"https://openalex.org/keywords/householder-reflections","display_name":"Householder reflections","score":0.8650000095367432},{"id":"https://openalex.org/keywords/transition-matrix","display_name":"transition matrix","score":0.8460000157356262},{"id":"https://openalex.org/keywords/long-term-dependencies","display_name":"long-range dependencies","score":0.8659999966621399},{"id":"https://openalex.org/keywords/vanishing-and-exploding-gradients","display_name":"vanishing and exploding gradients","score":0.8180000185966492},{"id":"https://openalex.org/keywords/time-series-forecasting","display_name":"time series forecasting","score":0.6650000214576721},{"id":"https://openalex.org/keywords/stochastic-gradient-descent","display_name":"stochastic gradient descent","score":0.5669999718666077},{"id":"https://openalex.org/keywords/computational-complexity","display_name":"computational complexity","score":0.6489999890327454},{"id":"https://openalex.org/keywords/orthogonal-matrices","display_name":"orthogonal matrices","score":0.375},{"id":"https://openalex.org/keywords/time-series-classification","display_name":"time series classification","score":0.3319999873638153},{"id":"https://openalex.org/keywords/time-series-regression","display_name":"time series regression","score":0.5149999856948853},{"id":"https://openalex.org/keywords/time-series-clustering","display_name":"time series clustering","score":0.5130000114440918},{"id":"https://openalex.org/keywords/time-series-segmentation","display_name":"time series segmentation","score":0.5180000066757202},{"id":"https://openalex.org/keywords/time-series-summarization","display_name":"time series summarization","score":0.5099999904632568},{"id":"https://openalex.org/keywords/time-series-anomaly-detection","display_name":"time series anomaly detection","score":0.46299999952316284}],"concepts":[{"id":"https://openalex.org/C67820243","wikidata":"https://www.wikidata.org/wiki/Q179164","display_name":"Unitary state","level":2,"score":0.6976763010025024},{"id":"https://openalex.org/C147168706","wikidata":"https://www.wikidata.org/wiki/Q1457734","display_name":"Recurrent neural network","level":3,"score":0.6867921352386475},{"id":"https://openalex.org/C2776036281","wikidata":"https://www.wikidata.org/wiki/Q48769818","display_name":"Constraint (computer-aided design)","level":2,"score":0.6503464579582214},{"id":"https://openalex.org/C49555168","wikidata":"https://www.wikidata.org/wiki/Q176583","display_name":"Stochastic matrix","level":3,"score":0.5554020404815674},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.5546073317527771},{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.5301008820533752},{"id":"https://openalex.org/C206688291","wikidata":"https://www.wikidata.org/wiki/Q7617819","display_name":"Stochastic gradient descent","level":3,"score":0.5254848599433899},{"id":"https://openalex.org/C106487976","wikidata":"https://www.wikidata.org/wiki/Q685816","display_name":"Matrix (chemical analysis)","level":2,"score":0.5077205896377563},{"id":"https://openalex.org/C126255220","wikidata":"https://www.wikidata.org/wiki/Q141495","display_name":"Mathematical optimization","level":1,"score":0.48717162013053894},{"id":"https://openalex.org/C11413529","wikidata":"https://www.wikidata.org/wiki/Q8366","display_name":"Algorithm","level":1,"score":0.48368531465530396},{"id":"https://openalex.org/C22629506","wikidata":"https://www.wikidata.org/wiki/Q7602993","display_name":"State-transition matrix","level":4,"score":0.46517395973205566},{"id":"https://openalex.org/C92207270","wikidata":"https://www.wikidata.org/wiki/Q939253","display_name":"Matrix norm","level":3,"score":0.4605743885040283},{"id":"https://openalex.org/C153258448","wikidata":"https://www.wikidata.org/wiki/Q1199743","display_name":"Gradient descent","level":3,"score":0.4591725170612335},{"id":"https://openalex.org/C191795146","wikidata":"https://www.wikidata.org/wiki/Q3878446","display_name":"Norm (philosophy)","level":2,"score":0.45903730392456055},{"id":"https://openalex.org/C44292817","wikidata":"https://www.wikidata.org/wiki/Q333871","display_name":"Orthogonal matrix","level":3,"score":0.4363476037979126},{"id":"https://openalex.org/C28826006","wikidata":"https://www.wikidata.org/wiki/Q33521","display_name":"Applied mathematics","level":1,"score":0.3531933128833771},{"id":"https://openalex.org/C33923547","wikidata":"https://www.wikidata.org/wiki/Q395","display_name":"Mathematics","level":0,"score":0.35239177942276},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.27343207597732544},{"id":"https://openalex.org/C54848796","wikidata":"https://www.wikidata.org/wiki/Q339011","display_name":"Symmetric matrix","level":3,"score":0.25371813774108887},{"id":"https://openalex.org/C187064257","wikidata":"https://www.wikidata.org/wiki/Q3306808","display_name":"Orthogonal basis","level":2,"score":0.1286085844039917},{"id":"https://openalex.org/C119857082","wikidata":"https://www.wikidata.org/wiki/Q2539","display_name":"Machine learning","level":1,"score":0.11751681566238403},{"id":"https://openalex.org/C158693339","wikidata":"https://www.wikidata.org/wiki/Q190524","display_name":"Eigenvalues and eigenvectors","level":2,"score":0.08268347382545471},{"id":"https://openalex.org/C159985019","wikidata":"https://www.wikidata.org/wiki/Q181790","display_name":"Composite material","level":1,"score":0.0},{"id":"https://openalex.org/C62520636","wikidata":"https://www.wikidata.org/wiki/Q944","display_name":"Quantum mechanics","level":1,"score":0.0},{"id":"https://openalex.org/C199539241","wikidata":"https://www.wikidata.org/wiki/Q7748","display_name":"Law","level":1,"score":0.0},{"id":"https://openalex.org/C2524010","wikidata":"https://www.wikidata.org/wiki/Q8087","display_name":"Geometry","level":1,"score":0.0},{"id":"https://openalex.org/C192562407","wikidata":"https://www.wikidata.org/wiki/Q228736","display_name":"Materials science","level":0,"score":0.0},{"id":"https://openalex.org/C17744445","wikidata":"https://www.wikidata.org/wiki/Q36442","display_name":"Political science","level":0,"score":0.0},{"id":"https://openalex.org/C121332964","wikidata":"https://www.wikidata.org/wiki/Q413","display_name":"Physics","level":0,"score":0.0},{"id":"https://openalex.org/C98763669","wikidata":"https://www.wikidata.org/wiki/Q176645","display_name":"Markov chain","level":2,"score":0.0}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:jupiter.its.unimelb.edu.au:11343/192906","doi":null,"is_oa":true,"landing_page_url":"http://hdl.handle.net/11343/192906","pdf_url":null,"source":{"id":"https://openalex.org/S4377196259","display_name":"Minerva Access (University of Melbourne)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I165779595","host_organization_name":"The University of Melbourne","host_organization_lineage":["https://openalex.org/I165779595"],"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":"Masters Research thesis"},{"id":"mag:2963992014","doi":null,"is_oa":false,"landing_page_url":"http://proceedings.mlr.press/v70/mhammedi17a/mhammedi17a.pdf","pdf_url":null,"source":{"id":"https://openalex.org/S4306419644","display_name":"International Conference on Machine Learning","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":null,"host_organization_name":null,"host_organization_lineage":[],"host_organization_lineage_names":[],"type":"conference"},"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":"International Conference on Machine Learning","raw_type":null}],"best_oa_location":{"id":"pmh:oai:jupiter.its.unimelb.edu.au:11343/192906","doi":null,"is_oa":true,"landing_page_url":"http://hdl.handle.net/11343/192906","pdf_url":null,"source":{"id":"https://openalex.org/S4377196259","display_name":"Minerva Access (University of Melbourne)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I165779595","host_organization_name":"The University of Melbourne","host_organization_lineage":["https://openalex.org/I165779595"],"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":"Masters Research thesis"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"grobid_xml":true,"pdf":true},"content_urls":{"pdf":"https://content.openalex.org/works/W2963992014.pdf","grobid_xml":"https://content.openalex.org/works/W2963992014.grobid-xml"},"referenced_works_count":51,"referenced_works":["https://openalex.org/W4919037","https://openalex.org/W104184427","https://openalex.org/W581956982","https://openalex.org/W1408639475","https://openalex.org/W1484210532","https://openalex.org/W1525783482","https://openalex.org/W1533861849","https://openalex.org/W1556310567","https://openalex.org/W1598796236","https://openalex.org/W1800356822","https://openalex.org/W1810943226","https://openalex.org/W1847088711","https://openalex.org/W1850742715","https://openalex.org/W2024450772","https://openalex.org/W2040870580","https://openalex.org/W2064675550","https://openalex.org/W2086283032","https://openalex.org/W2090563320","https://openalex.org/W2101927907","https://openalex.org/W2107878631","https://openalex.org/W2110798204","https://openalex.org/W2126813895","https://openalex.org/W2133319764","https://openalex.org/W2136655611","https://openalex.org/W2137983211","https://openalex.org/W2161663128","https://openalex.org/W2167823677","https://openalex.org/W2167839676","https://openalex.org/W2176412452","https://openalex.org/W2201092681","https://openalex.org/W2234163201","https://openalex.org/W2251189452","https://openalex.org/W2312434537","https://openalex.org/W2338522602","https://openalex.org/W2384495648","https://openalex.org/W2401137308","https://openalex.org/W2474920236","https://openalex.org/W2493026224","https://openalex.org/W2547718140","https://openalex.org/W2557270725","https://openalex.org/W2587753047","https://openalex.org/W2901851926","https://openalex.org/W2964121744","https://openalex.org/W3148186152","https://openalex.org/W4205130185","https://openalex.org/W4233045210","https://openalex.org/W4236047370","https://openalex.org/W4394663752","https://openalex.org/W6907861883","https://openalex.org/W6996569244","https://openalex.org/W7009541382"],"related_works":["https://openalex.org/W2943592883","https://openalex.org/W7071098563","https://openalex.org/W2914344989","https://openalex.org/W3016617846","https://openalex.org/W3192210946","https://openalex.org/W1483397016","https://openalex.org/W4321161984","https://openalex.org/W2740137140","https://openalex.org/W3154976557","https://openalex.org/W3131340183"],"abstract_inverted_index":{"In":[0],"machine":[1],"learning,":[2],"Recurrent":[3],"Neural":[4],"Networks":[5],"(RNNs)":[6],"have":[7,100,146],"been":[8,101,389],"successfully":[9],"used":[10,247],"in":[11,54,70,248,288,294,382,391,407],"many":[12,45],"applications.They":[13],"are":[14,56,58,90,172,306],"particularly":[15],"well":[16],"suited":[17],"for":[18],"problems":[19,338],"involving":[20,49],"time-series.This":[21],"is":[22,41,74,79,122,185,224,259,271,323,360,402],"because":[23],"RNNs":[24,62,73],"process":[25],"input":[26,96,311],"sequences":[27,71,97],"one":[28,283],"element":[29],"at":[30],"a":[31,76,186,205,227],"time,":[32],"and":[33,85,232,304,313,414],"thus,":[34],"the":[35,59,83,93,95,109,132,135,153,156,161,179,209,222,239,249,254,262,265,268,275,289,307,310,314,317,324,329,345,349,356,365,373,383,392,400],"chronological":[36],"order":[37],"of":[38,134,155,189,208,216,242,256,264,278,309,316,328,375,377,410],"these":[39,67],"elements":[40],"taken":[42],"into":[43],"account.In":[44],"practical":[46],"prediction":[47],"tasks":[48],"time-series,":[50],"long-term":[51,68,340],"time":[52,295,321],"dependencies":[53,69],"data":[55],"crucial.These":[57],"features":[60],"that":[61,119,177,221,327,344,363],"need":[63],"to":[64,82,103,112,125,182,198,261,273,355,361,394],"learn.However,":[65],"learning":[66],"using":[72,167],"still":[75],"major":[77],"challenge.This":[78],"mainly":[80],"due":[81,386],"exploding":[84],"vanishing":[86],"gradient":[87,169,285],"problems,":[88],"which":[89,117,212],"more":[91],"prominent":[92],"longer":[94],"are.Recent":[98],"methods":[99,144],"suggested":[102],"solve":[104],"this":[105],"problem":[106],"by":[107,237],"constraining":[108,178],"transition":[110,180,210,269,350],"matrix":[111,181,211,223,270,351],"be":[113,183,196,235,292],"unitary":[114,184,357],"during":[115],"training,":[116],"ensures":[118],"its":[120],"norm":[121,133],"exactly":[123],"equal":[124,260],"one.This":[126],"sets":[127],"an":[128,190,217],"upper":[129],"bound":[130],"on":[131,337,348],"back-propagated":[136],"gradients,":[137],"preventing":[138],"them":[139],"from":[140],"growing":[141],"exponentially.However,":[142],"existing":[143],"either":[145],"limited":[147],"expressiveness":[148,233],"or":[149],"scale":[150],"poorly":[151],"with":[152,160,200,339],"size":[154,263,315],"network":[157],"when":[158,166,253],"compared":[159],"simple":[162,330],"RNN":[163,218],"case,":[164,291],"especially":[165],"stochastic":[168,284],"descent.Our":[170],"contributions":[171],"as":[173,326],"follows.We":[174],"first":[175],"show":[176],"special":[187],"case":[188],"orthogonal":[191,279,346],"constraint.Therefore,":[192],"it":[193],"may":[194],"not":[195],"necessary":[197],"work":[199,371],"complex-valued":[201],"matrices.Then":[202],"we":[203,332],"present":[204],"new":[206,250,335],"parametrisation":[207,336],"allows":[213],"efficient":[214],"training":[215],"while":[219],"ensuring":[220],"always":[225],"orthogonal.Furthermore,":[226],"good":[228],"trade-off":[229],"between":[230],"speed":[231],"can":[234],"chosen":[236],"selecting":[238],"right":[240],"number":[241,255],"reflection":[243,257],"vectors":[244,246],"-the":[245],"parametrisation.In":[251],"particular,":[252],"vector":[258],"hidden":[266,318],"layer":[267,319],"allowed":[272],"span":[274],"full":[276],"set":[277],"matrices.Using":[280],"our":[281,334],"approach,":[282],"step":[286],"can,":[287],"worst":[290],"performed":[293],"comiii":[296],"plexity":[297],"O(T":[298],"n":[299,305],"2":[300],"),":[301],"where":[302,380],"T":[303],"length":[308],"sequence":[312],"respectively.This":[320],"complexity":[322],"same":[325],"RNN.Finally,":[331],"test":[333],"dependencies.Our":[341],"results":[342],"suggest":[343],"constraint":[347],"has":[352,388],"similar":[353],"benefits":[354],"constraint.":[358],"DeclarationThis":[359],"certify":[362],"1.":[364],"thesis":[366,401],"comprises":[367],"only":[368],"my":[369],"original":[370],"towards":[372],"degree":[374],"Master":[376],"Philosophy":[378],"except":[379],"indicated":[381],"Preface,":[384],"2.":[385],"acknowledgment":[387],"made":[390],"text":[393],"all":[395],"other":[396],"material":[397],"used,":[398],"3.":[399],"fewer":[403],"than":[404],"40,000":[405],"words":[406],"length,":[408],"exclusive":[409],"tables,":[411],"maps,":[412],"bibliographies":[413],"appendices.":[415]},"cited_by_api_url":"https://api.openalex.org/works?filter=cites:W2963992014","counts_by_year":[{"year":2026,"cited_by_count":20},{"year":2025,"cited_by_count":10},{"year":2024,"cited_by_count":17},{"year":2023,"cited_by_count":13},{"year":2022,"cited_by_count":13},{"year":2021,"cited_by_count":25},{"year":2020,"cited_by_count":22},{"year":2019,"cited_by_count":12},{"year":2018,"cited_by_count":5},{"year":2017,"cited_by_count":2}],"updated_date":"2026-11-16T12:05:36.265796","created_date":"2019-07-30T00:00:00"}
