Matches in SemOpenAlex for { <https://semopenalex.org/work/W4309795870> ?p ?o ?g. }
Showing items 1 to 73 of
73
with 100 items per page.
- W4309795870 abstract "We study the learning ability of linear recurrent neural networks with gradient descent. We prove the first theoretical guarantee on linear RNNs with Gradient Descent to learn any stable linear dynamic system. We show that despite the non-convexity of the optimization loss if the width of the RNN is large enough (and the required width in hidden layers does not rely on the length of the input sequence), a linear RNN can provably learn any stable linear dynamic system with the sample and time complexity polynomial in $frac{1}{1-rho_C}$ where $rho_C$ is roughly the spectral radius of the stable system. Our results provide the first theoretical guarantee to learn a linear RNN and demonstrate how can the recurrent structure help to learn a dynamic system." @default.
- W4309795870 created "2022-11-29" @default.
- W4309795870 creator A5059850766 @default.
- W4309795870 creator A5063439393 @default.
- W4309795870 creator A5072854024 @default.
- W4309795870 creator A5076449077 @default.
- W4309795870 date "2022-11-18" @default.
- W4309795870 modified "2023-09-27" @default.
- W4309795870 title "Can Gradient Descent Provably Learn Linear Dynamic Systems?" @default.
- W4309795870 doi "https://doi.org/10.48550/arxiv.2211.10582" @default.
- W4309795870 hasPublicationYear "2022" @default.
- W4309795870 type Work @default.
- W4309795870 citedByCount "0" @default.
- W4309795870 crossrefType "posted-content" @default.
- W4309795870 hasAuthorship W4309795870A5059850766 @default.
- W4309795870 hasAuthorship W4309795870A5063439393 @default.
- W4309795870 hasAuthorship W4309795870A5072854024 @default.
- W4309795870 hasAuthorship W4309795870A5076449077 @default.
- W4309795870 hasBestOaLocation W43097958701 @default.
- W4309795870 hasConcept C106159729 @default.
- W4309795870 hasConcept C11413529 @default.
- W4309795870 hasConcept C114275822 @default.
- W4309795870 hasConcept C134306372 @default.
- W4309795870 hasConcept C147168706 @default.
- W4309795870 hasConcept C153258448 @default.
- W4309795870 hasConcept C154945302 @default.
- W4309795870 hasConcept C162324750 @default.
- W4309795870 hasConcept C2778112365 @default.
- W4309795870 hasConcept C2778445095 @default.
- W4309795870 hasConcept C311688 @default.
- W4309795870 hasConcept C33923547 @default.
- W4309795870 hasConcept C41008148 @default.
- W4309795870 hasConcept C50644808 @default.
- W4309795870 hasConcept C54355233 @default.
- W4309795870 hasConcept C6802819 @default.
- W4309795870 hasConcept C72134830 @default.
- W4309795870 hasConcept C86803240 @default.
- W4309795870 hasConcept C90119067 @default.
- W4309795870 hasConceptScore W4309795870C106159729 @default.
- W4309795870 hasConceptScore W4309795870C11413529 @default.
- W4309795870 hasConceptScore W4309795870C114275822 @default.
- W4309795870 hasConceptScore W4309795870C134306372 @default.
- W4309795870 hasConceptScore W4309795870C147168706 @default.
- W4309795870 hasConceptScore W4309795870C153258448 @default.
- W4309795870 hasConceptScore W4309795870C154945302 @default.
- W4309795870 hasConceptScore W4309795870C162324750 @default.
- W4309795870 hasConceptScore W4309795870C2778112365 @default.
- W4309795870 hasConceptScore W4309795870C2778445095 @default.
- W4309795870 hasConceptScore W4309795870C311688 @default.
- W4309795870 hasConceptScore W4309795870C33923547 @default.
- W4309795870 hasConceptScore W4309795870C41008148 @default.
- W4309795870 hasConceptScore W4309795870C50644808 @default.
- W4309795870 hasConceptScore W4309795870C54355233 @default.
- W4309795870 hasConceptScore W4309795870C6802819 @default.
- W4309795870 hasConceptScore W4309795870C72134830 @default.
- W4309795870 hasConceptScore W4309795870C86803240 @default.
- W4309795870 hasConceptScore W4309795870C90119067 @default.
- W4309795870 hasLocation W43097958701 @default.
- W4309795870 hasOpenAccess W4309795870 @default.
- W4309795870 hasPrimaryLocation W43097958701 @default.
- W4309795870 hasRelatedWork W2054802127 @default.
- W4309795870 hasRelatedWork W2158982621 @default.
- W4309795870 hasRelatedWork W2440206734 @default.
- W4309795870 hasRelatedWork W2545595979 @default.
- W4309795870 hasRelatedWork W2594213702 @default.
- W4309795870 hasRelatedWork W3130927411 @default.
- W4309795870 hasRelatedWork W3216774278 @default.
- W4309795870 hasRelatedWork W4225854624 @default.
- W4309795870 hasRelatedWork W4309795870 @default.
- W4309795870 hasRelatedWork W4375851942 @default.
- W4309795870 isParatext "false" @default.
- W4309795870 isRetracted "false" @default.
- W4309795870 workType "article" @default.