{"id":"https://openalex.org/W2748193545","doi":null,"title":"A Novel Mathematical Framework for the Analysis of Neural Networks","display_name":"A Novel Mathematical Framework for the Analysis of Neural Networks","publication_year":2017,"publication_date":"2017-08-22","ids":{"openalex":"https://openalex.org/W2748193545","mag":"2748193545"},"language":"en","primary_location":{"id":"pmh:oai:uwspace.uwaterloo.ca:10012/12173","doi":null,"is_oa":true,"landing_page_url":"http://hdl.handle.net/10012/12173","pdf_url":"https://uwspace.uwaterloo.ca/bitstreams/240b59bb-39f5-4e9b-860e-6754e2d2319f/download","source":{"id":"https://openalex.org/S4306401661","display_name":"UWSpace (University of Waterloo)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I151746483","host_organization_name":"University of Waterloo","host_organization_lineage":["https://openalex.org/I151746483"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"Master Thesis"},"volume":null,"type":"dissertation","is_traditional":true,"indexed_in":[],"open_access":{"is_oa":true,"oa_status":"green","oa_url":"https://uwspace.uwaterloo.ca/bitstreams/240b59bb-39f5-4e9b-860e-6754e2d2319f/download","any_repository_has_fulltext":true},"authorships":[{"author_position":"first","author":{"id":"https://openalex.org/A5006499420","display_name":"Anthony L. Caterini","orcid":"https://orcid.org/0000-0002-0758-9562","observed_orcids":["https://orcid.org/0000-0002-0758-9562"]},"institutions":[],"countries":[],"is_corresponding":true,"raw_author_name":"Anthony L. Caterini","raw_affiliation_strings":[],"raw_orcid":null,"affiliations":[]}],"institutions":[],"countries_distinct_count":0,"institutions_distinct_count":0,"corresponding_author_ids":["https://openalex.org/A5006499420"],"corresponding_institution_ids":[],"apc_list":null,"apc_paid":null,"fwci":null,"has_fulltext":true,"cited_by_count":6,"citation_normalized_percentile":null,"cited_by_percentile_year":null,"biblio":{"volume":null,"issue":null,"first_page":null,"last_page":null},"is_retracted":false,"is_paratext":false,"is_xpac":false,"primary_topic":{"id":"https://openalex.org/T10320","display_name":"Neural Networks and Applications","score":0.9980000257492065,"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":0.9980000257492065,"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/T11612","display_name":"Stochastic Gradient Optimization Techniques","score":0.0010999999940395355,"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/T12676","display_name":"Machine Learning and ELM","score":0.00019999999494757503,"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/deep-neural-network","display_name":"deep neural network","score":0.8799999952316284},{"id":"https://openalex.org/keywords/vector-spaces","display_name":"vector spaces","score":0.7039999961853027},{"id":"https://openalex.org/keywords/inner-product-spaces","display_name":"inner product spaces","score":0.7120000123977661},{"id":"https://openalex.org/keywords/gradient-descent","display_name":"gradient descent","score":0.9190000295639038},{"id":"https://openalex.org/keywords/backpropagation","display_name":"backpropagation","score":0.8769999742507935},{"id":"https://openalex.org/keywords/multilayer-perceptron","display_name":"multilayer perceptron","score":0.9599999785423279},{"id":"https://openalex.org/keywords/convolutional-neural-networks","display_name":"convolutional neural networks","score":1.0},{"id":"https://openalex.org/keywords/deep-autoencoder","display_name":"deep autoencoder","score":0.8690000176429749},{"id":"https://openalex.org/keywords/recurrent-neural-networks","display_name":"recurrent neural networks","score":0.9990000128746033},{"id":"https://openalex.org/keywords/weight-sharing","display_name":"parameter sharing","score":0.41200000047683716},{"id":"https://openalex.org/keywords/loss-functions","display_name":"loss functions","score":0.3919999897480011},{"id":"https://openalex.org/keywords/neural-network-architecture","display_name":"neural network architecture","score":0.3190000057220459},{"id":"https://openalex.org/keywords/matrix-calculus","display_name":"matrix calculus","score":0.3569999933242798},{"id":"https://openalex.org/keywords/natural-language-processing","display_name":"natural language processing","score":0.33799999952316284},{"id":"https://openalex.org/keywords/machine-translation","display_name":"machine translation","score":0.3569999933242798}],"concepts":[{"id":"https://openalex.org/C50644808","wikidata":"https://www.wikidata.org/wiki/Q192776","display_name":"Artificial neural network","level":2,"score":0.5063823461532593},{"id":"https://openalex.org/C41008148","wikidata":"https://www.wikidata.org/wiki/Q21198","display_name":"Computer science","level":0,"score":0.48858270049095154},{"id":"https://openalex.org/C154945302","wikidata":"https://www.wikidata.org/wiki/Q11660","display_name":"Artificial intelligence","level":1,"score":0.3708624541759491},{"id":"https://openalex.org/C2522767166","wikidata":"https://www.wikidata.org/wiki/Q2374463","display_name":"Data science","level":1,"score":0.3271111845970154}],"mesh":[],"locations_count":2,"locations":[{"id":"pmh:oai:uwspace.uwaterloo.ca:10012/12173","doi":null,"is_oa":true,"landing_page_url":"http://hdl.handle.net/10012/12173","pdf_url":"https://uwspace.uwaterloo.ca/bitstreams/240b59bb-39f5-4e9b-860e-6754e2d2319f/download","source":{"id":"https://openalex.org/S4306401661","display_name":"UWSpace (University of Waterloo)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I151746483","host_organization_name":"University of Waterloo","host_organization_lineage":["https://openalex.org/I151746483"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"Master Thesis"},{"id":"mag:2748193545","doi":null,"is_oa":false,"landing_page_url":"https://uwspace.uwaterloo.ca/handle/10012/12173","pdf_url":null,"source":null,"license":null,"license_id":null,"version":null,"is_accepted":false,"is_published":false,"raw_source_name":null,"raw_type":null}],"best_oa_location":{"id":"pmh:oai:uwspace.uwaterloo.ca:10012/12173","doi":null,"is_oa":true,"landing_page_url":"http://hdl.handle.net/10012/12173","pdf_url":"https://uwspace.uwaterloo.ca/bitstreams/240b59bb-39f5-4e9b-860e-6754e2d2319f/download","source":{"id":"https://openalex.org/S4306401661","display_name":"UWSpace (University of Waterloo)","issn_l":null,"issn":null,"is_oa":false,"is_in_doaj":false,"is_core":false,"listed_in":[],"host_organization":"https://openalex.org/I151746483","host_organization_name":"University of Waterloo","host_organization_lineage":["https://openalex.org/I151746483"],"host_organization_lineage_names":[],"type":"repository"},"license":null,"license_id":null,"version":"submittedVersion","is_accepted":false,"is_published":false,"raw_source_name":"","raw_type":"Master Thesis"},"sustainable_development_goals":[],"study_designs":[],"awards":[],"funders":[],"has_content":{"pdf":true,"grobid_xml":true},"content_urls":{"pdf":"https://content.openalex.org/works/W2748193545.pdf","grobid_xml":"https://content.openalex.org/works/W2748193545.grobid-xml"},"referenced_works_count":71,"referenced_works":["https://openalex.org/W20283819","https://openalex.org/W117096852","https://openalex.org/W194249466","https://openalex.org/W197865394","https://openalex.org/W581956982","https://openalex.org/W1525783482","https://openalex.org/W1563017489","https://openalex.org/W1572063013","https://openalex.org/W1597009674","https://openalex.org/W1597220786","https://openalex.org/W1686810756","https://openalex.org/W1810943226","https://openalex.org/W1853291804","https://openalex.org/W1889624880","https://openalex.org/W1895577753","https://openalex.org/W1924770834","https://openalex.org/W1981530182","https://openalex.org/W1988115241","https://openalex.org/W1995341919","https://openalex.org/W2005708641","https://openalex.org/W2011662639","https://openalex.org/W2016589492","https://openalex.org/W2040870580","https://openalex.org/W2042492924","https://openalex.org/W2064675550","https://openalex.org/W2076063813","https://openalex.org/W2086234878","https://openalex.org/W2099471712","https://openalex.org/W2100115174","https://openalex.org/W2100495367","https://openalex.org/W2103496339","https://openalex.org/W2110798204","https://openalex.org/W2111494971","https://openalex.org/W2112796928","https://openalex.org/W2114471683","https://openalex.org/W2119169789","https://openalex.org/W2131774270","https://openalex.org/W2132424367","https://openalex.org/W2133319764","https://openalex.org/W2136036867","https://openalex.org/W2136922672","https://openalex.org/W2144513243","https://openalex.org/W2145339207","https://openalex.org/W2146989110","https://openalex.org/W2154579312","https://openalex.org/W2157331557","https://openalex.org/W2160815625","https://openalex.org/W2161388792","https://openalex.org/W2173520492","https://openalex.org/W2176412452","https://openalex.org/W2184271171","https://openalex.org/W2194775991","https://openalex.org/W2218318129","https://openalex.org/W2252143850","https://openalex.org/W2257979135","https://openalex.org/W2267573953","https://openalex.org/W2321916036","https://openalex.org/W2432004435","https://openalex.org/W2511261638","https://openalex.org/W2513846677","https://openalex.org/W2523246573","https://openalex.org/W2528509685","https://openalex.org/W2581485081","https://openalex.org/W2766736793","https://openalex.org/W2806070179","https://openalex.org/W2919115771","https://openalex.org/W2950635152","https://openalex.org/W2962742960","https://openalex.org/W2993845287","https://openalex.org/W3037567775","https://openalex.org/W4300457772"],"related_works":["https://openalex.org/W2528509685","https://openalex.org/W2790567079","https://openalex.org/W2792247102","https://openalex.org/W4414200602","https://openalex.org/W2804742100","https://openalex.org/W2789585573","https://openalex.org/W4386728290","https://openalex.org/W7160509159","https://openalex.org/W7161693758","https://openalex.org/W2513846677"],"abstract_inverted_index":{"Over":[0],"the":[1,45,124,145,181,195,217,223,231,266,270,275,292,295,314,338,348,367,384,387,399,411,431,435],"past":[2],"decade,":[3],"Deep":[4,349],"Neural":[5,316,396],"Networks":[6,397],"(DNNs)":[7],"have":[8],"become":[9],"very":[10],"popular":[11],"models":[12,31],"for":[13,78,208,309,394],"processing":[14],"large":[15],"amounts":[16],"of":[17,20,28,115,126,178,212,269,298,383,410,434],"data":[18],"because":[19],"their":[21,155,164],"successful":[22],"application":[23],"in":[24,44,198,221,235,286],"a":[25,74,127,174,236,242,260,305,392,425],"wide":[26],"variety":[27],"fields.":[29],"These":[30],"are":[32,58,187,202,225,355,363,405],"layered,":[33],"often":[34],"containing":[35],"parametrized":[36],"linear":[37],"and":[38,82,147,189,238,300,327,413,429],"non-linear":[39],"transformations":[40,328],"at":[41,329,359],"each":[42,330,360],"layer":[43],"network.":[46,130,271],"At":[47],"this":[48,62,70,85,372],"point,":[49],"however,":[50,334],"we":[51,64,72,100,185,228,249,278,345,380,415],"do":[52],"not":[53,356],"rigorously":[54,111],"understand":[55],"why":[56],"DNNs":[57,179],"so":[59],"effective.":[60],"In":[61],"thesis,":[63],"explore":[65],"one":[66,210],"way":[67],"to":[68,90,106,121,137,153,204,259,281,303,365,370,390,420],"approach":[69],"problem:":[71],"develop":[73,194,391],"generic":[75,128,196,276,339,368,426],"mathematical":[76,104,182],"framework":[77,86,167,197,369,393],"representing":[79],"neural":[80,93,107,129,139],"networks,":[81],"demonstrate":[83,251],"how":[84,302],"can":[87,110,229],"be":[88],"used":[89],"represent":[91,230,313],"specific":[92,283,432],"network":[94,284],"architectures.":[95],"\\n":[96,191,272,376],"\\nIn":[97,377],"chapter":[98,199,287,378],"1,":[99],"start":[101,290],"by":[102],"exploring":[103],"contributions":[105],"networks.":[108],"We":[109,131,201,289,311,362,423,438],"explain":[112],"some":[113,161,382,417],"properties":[114],"DNNs,":[116],"but":[117],"these":[118,171],"results":[119,385],"fail":[120],"fully":[122],"describe":[123,205,421,424],"mechanics":[125],"also":[132,250],"note":[133],"that":[134,184,252,344,354],"most":[135],"approaches":[136],"describing":[138],"networks":[140],"rely":[141],"upon":[142],"breaking":[143],"down":[144],"parameters":[146,224,353,404],"inputs":[148],"into":[149,163,337],"scalars,":[150],"as":[151,374],"opposed":[152],"referencing":[154],"underlying":[156],"vector":[157],"spaces,":[158,172,324,326],"which":[159,222,319,351],"adds":[160],"awkwardness":[162],"analysis.":[165],"Our":[166],"strictly":[168],"operates":[169],"over":[170,216,443],"affording":[173],"more":[175,261,321],"natural":[176],"description":[177],"once":[180],"objects":[183],"use":[186,381],"well-defined":[188],"understood.":[190],"\\nWe":[192],"then":[193,312,430],"3.":[200],"able":[203,364],"an":[206],"algorithm":[207],"calculating":[209],"step":[211,234,308],"gradient":[213,256,306],"descent":[214,307],"directly":[215,442],"inner":[218,444],"product":[219,445],"space":[220],"defined.":[226],"Also,":[227],"error":[232],"backpropagation":[233],"concise":[237],"compact":[239],"form.":[240],"Besides":[241],"standard":[243],"squared":[244],"loss":[245,263],"or":[246],"cross-entropy":[247],"loss,":[248],"our":[253],"framework,":[254,277],"including":[255],"calculation,":[257],"extends":[258],"complex":[262],"function":[264],"involving":[265],"first":[267],"derivative":[268],"\\nAfter":[273],"developing":[274],"apply":[279],"it":[280],"three":[282],"examples":[285],"4.":[288],"with":[291],"Multilayer":[293],"Perceptron,":[294],"simplest":[296],"type":[297],"DNN,":[299],"show":[301],"generate":[304],"it.":[310],"Convolutional":[315],"Network":[317],"(CNN),":[318],"contains":[320],"complicated":[322],"input":[323],"parameter":[325],"layer.":[331,361],"The":[332,341,403],"CNN,":[333],"still":[335],"fits":[336],"framework.":[340],"last":[342],"structure":[343],"consider":[346],"is":[347],"Auto-Encoder,":[350],"has":[352],"completely":[357],"independent":[358],"extend":[366],"handle":[371],"case":[373,433],"well.":[375],"5,":[379],"from":[386],"previous":[388],"chapters":[389],"Recurrent":[395],"(RNNs),":[398],"sequence-parsing":[400],"DNN":[401],"architecture.":[402],"shared":[406],"across":[407],"all":[408],"layers":[409],"network,":[412],"thus":[414],"require":[416],"additional":[418],"machinery":[419],"RNNs.":[422],"RNN":[427],"first,":[428],"vanilla":[436],"RNN.":[437],"again":[439],"compute":[440],"gradients":[441],"spaces.":[446]},"counts_by_year":[{"year":2025,"cited_by_count":1},{"year":2024,"cited_by_count":1},{"year":2021,"cited_by_count":1},{"year":2020,"cited_by_count":2},{"year":2019,"cited_by_count":1}],"updated_date":"2026-11-07T11:25:07.637180","created_date":"2017-08-31T00:00:00"}