ASSOCIATIVE LONG SHORT-TERM MEMORY NEURAL NETWORK LAYERS

Systems, methods, and apparatus, including computer programs encoded on a computer storage medium, related to associative long short-term memory (LSTM) neural network layers configured to maintain N copies of an internal state for the associative LSTM layer, N being an integer greater than one. In o...

Full description

Saved in:
Bibliographic Details
Main Authors DANIHELKA, Ivo, KALCHBRENNER, Nal Emmerich, WAYNE, Gregory Duncan, URIA-MARTINEZ, Benigno, GRAVES, Alexander Benjamin
Format Patent
LanguageEnglish
French
Published 10.08.2017
Subjects
Online AccessGet full text

Cover

Loading…
Abstract Systems, methods, and apparatus, including computer programs encoded on a computer storage medium, related to associative long short-term memory (LSTM) neural network layers configured to maintain N copies of an internal state for the associative LSTM layer, N being an integer greater than one. In one aspect, a system includes a recurrent neural network including an associative LSTM layer, wherein the associative LSTM layer is configured to, for each time step, receive a layer input, update each of the N copies of the internal state using the layer input for the time step and a layer output generated by the associative LSTM layer for a preceding time step, and generate a layer output for the time step using the N updated copies of the internal state. L'invention concerne des systèmes, des procédés et un appareil, comprenant des programmes informatiques codés sur un support d'enregistrement informatique, associés à des couches associatives de réseau neuronal de mémoire à long et court terme (LSTM) configurées pour conserver N copies d'un état interne de la couche associative de LSTM, N étant un nombre entier supérieur à un. Selon un aspect, un système comprend un réseau neuronal récursif comprenant une couche associative de LSTM, la couche associative de LSTM étant configurée pour, à chaque étape temporelle, recevoir une entrée de couche, mettre à jour chacune des N copies de l'état interne au moyen de l'entrée de couche de l'étape temporelle et d'une sortie de couche générée par la couche associative de LSTM lors d'une étape temporelle précédente, et générer une sortie de couche de l'étape temporelle au moyen des N copies mises à jour de l'état interne.
AbstractList Systems, methods, and apparatus, including computer programs encoded on a computer storage medium, related to associative long short-term memory (LSTM) neural network layers configured to maintain N copies of an internal state for the associative LSTM layer, N being an integer greater than one. In one aspect, a system includes a recurrent neural network including an associative LSTM layer, wherein the associative LSTM layer is configured to, for each time step, receive a layer input, update each of the N copies of the internal state using the layer input for the time step and a layer output generated by the associative LSTM layer for a preceding time step, and generate a layer output for the time step using the N updated copies of the internal state. L'invention concerne des systèmes, des procédés et un appareil, comprenant des programmes informatiques codés sur un support d'enregistrement informatique, associés à des couches associatives de réseau neuronal de mémoire à long et court terme (LSTM) configurées pour conserver N copies d'un état interne de la couche associative de LSTM, N étant un nombre entier supérieur à un. Selon un aspect, un système comprend un réseau neuronal récursif comprenant une couche associative de LSTM, la couche associative de LSTM étant configurée pour, à chaque étape temporelle, recevoir une entrée de couche, mettre à jour chacune des N copies de l'état interne au moyen de l'entrée de couche de l'étape temporelle et d'une sortie de couche générée par la couche associative de LSTM lors d'une étape temporelle précédente, et générer une sortie de couche de l'étape temporelle au moyen des N copies mises à jour de l'état interne.
Author WAYNE, Gregory Duncan
DANIHELKA, Ivo
KALCHBRENNER, Nal Emmerich
GRAVES, Alexander Benjamin
URIA-MARTINEZ, Benigno
Author_xml – fullname: DANIHELKA, Ivo
– fullname: KALCHBRENNER, Nal Emmerich
– fullname: WAYNE, Gregory Duncan
– fullname: URIA-MARTINEZ, Benigno
– fullname: GRAVES, Alexander Benjamin
BookMark eNrjYmDJy89L5WSwcAwO9nf2dAzxDHNV8PH3c1cI9vAPCtENcQ3yVfB19fUPilTwcw0NcvQBUiHh_kHeCj6Oka5BwTwMrGmJOcWpvFCam0HZzTXE2UM3tSA_PrW4IDE5NS-1JD7c38jA0NzQ2MzA3NzR0Jg4VQBARyuL
ContentType Patent
DBID EVB
DatabaseName esp@cenet
DatabaseTitleList
Database_xml – sequence: 1
  dbid: EVB
  name: esp@cenet
  url: http://worldwide.espacenet.com/singleLineSearch?locale=en_EP
  sourceTypes: Open Access Repository
DeliveryMethod fulltext_linktorsrc
Discipline Medicine
Chemistry
Sciences
Physics
DocumentTitleAlternate COUCHES ASSOCIATIVES DE RÉSEAU NEURONAL DE MÉMOIRE À LONG ET COURT TERME
ExternalDocumentID WO2017136077A1
GroupedDBID EVB
ID FETCH-epo_espacenet_WO2017136077A13
IEDL.DBID EVB
IngestDate Fri Jul 19 14:29:55 EDT 2024
IsOpenAccess true
IsPeerReviewed false
IsScholarly false
Language English
French
LinkModel DirectLink
MergedId FETCHMERGED-epo_espacenet_WO2017136077A13
Notes Application Number: WO2016US69505
OpenAccessLink https://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20170810&DB=EPODOC&CC=WO&NR=2017136077A1
ParticipantIDs epo_espacenet_WO2017136077A1
PublicationCentury 2000
PublicationDate 20170810
PublicationDateYYYYMMDD 2017-08-10
PublicationDate_xml – month: 08
  year: 2017
  text: 20170810
  day: 10
PublicationDecade 2010
PublicationYear 2017
RelatedCompanies GOOGLE INC
RelatedCompanies_xml – name: GOOGLE INC
Score 3.1070302
Snippet Systems, methods, and apparatus, including computer programs encoded on a computer storage medium, related to associative long short-term memory (LSTM) neural...
SourceID epo
SourceType Open Access Repository
SubjectTerms CALCULATING
COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
COMPUTING
COUNTING
PHYSICS
Title ASSOCIATIVE LONG SHORT-TERM MEMORY NEURAL NETWORK LAYERS
URI https://worldwide.espacenet.com/publicationDetails/biblio?FT=D&date=20170810&DB=EPODOC&locale=&CC=WO&NR=2017136077A1
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
link http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwfV3da8IwED_Efb5tbmMfbhQ2-lZWbWv0QUZt09XNNlLr15O0tYXBqLJ27N_fJejmk08hCRyXwOXul9z9AvBEUl2PE6IpJIs6ip5wDkgj42lTUZs0IyMzGrze2fNb7lh_mxmzCnxua2EET-iPIEdEi0rQ3ktxXq__L7FskVtZPMcfOLR6ccKuLW_QcYOgh1Nlu9elQ2YzS7YsxG2yH4i5htZSCTERKx1gIE14Ahid9HhdynrXqThncDhEeXl5DpU0r8GJtf17rQbH3ubJuwZHIkczKXBwY4fFBbQFG2PfDPsTKg2Y_yqNXBaECsamnuRRjwVzyafjwBxgE05Z8C4NzDkNRpfw6NDQchVUZvG39sWU7WquXUE1X-XpNUjNTqamWqS2cTf1eMnjjAjRnBajVSLIWt5AfZ-k2_3Td3DKu4rgfq1Dtfz6Tu_R-5bxg9i0X7a7gcI
link.rule.ids 230,309,783,888,25578,76884
linkProvider European Patent Office
linkToHtml http://utb.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwfV3dT8IwEL8Q_MA3RQ0q6hLN3hYH2-h4IGZsxSHbSsb4eiLb2BITM4jM-O97a0B54qlJL7m0l1zvfu3drwDPJFHVKCaKRNKwLalxwQGppUXZVKiTZqilWqPod3a9lj1W32farASfu14YzhP6w8kR0aNi9Pecn9fr_0ssi9dWbl6iD5xavfaCjiVu0XGDYISTRavboUNmMVM0TcRtoudzWUNpyYQYiJWOMMnWC6Z9OukWfSnr_aDSO4fjIerL8gsoJVkVKubu77UqnLrbJ-8qnPAazXiDk1s_3FyCztkY-0bQn1DBYd6bMLKZH0iYm7qCS13mzwWPjn3DwSGYMn8gOMac-qMreOrRwLQlXMzib--LKdtfuXIN5WyVJTUQmu1UTpRQ1tGaarQs8owQ0ZwSoVciyFreQP2QptvD4keo2IHrLJy-N7iDs0IkcR7YOpTzr-_kHiNxHj1wA_4C9J2Esg
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Apatent&rft.title=ASSOCIATIVE+LONG+SHORT-TERM+MEMORY+NEURAL+NETWORK+LAYERS&rft.inventor=DANIHELKA%2C+Ivo&rft.inventor=KALCHBRENNER%2C+Nal+Emmerich&rft.inventor=WAYNE%2C+Gregory+Duncan&rft.inventor=URIA-MARTINEZ%2C+Benigno&rft.inventor=GRAVES%2C+Alexander+Benjamin&rft.date=2017-08-10&rft.externalDBID=A1&rft.externalDocID=WO2017136077A1