Matches in SemOpenAlex for { <https://semopenalex.org/work/W3169482895> ?p ?o ?g. }
Showing items 1 to 83 of
83
with 100 items per page.
- W3169482895 endingPage "6769" @default.
- W3169482895 startingPage "6759" @default.
- W3169482895 abstract "The connection between training deep neural networks (DNNs) and optimal control theory (OCT) has attracted considerable attention as a principled tool of algorithmic design. Despite few attempts being made, they have been limited to architectures where the layer propagation resembles a Markovian dynamical system. This casts doubts on their flexibility to modern networks that heavily rely on non-Markovian dependencies between layers (e.g. skip connections in residual networks). In this work, we propose a novel dynamic game perspective by viewing each layer as a player in a dynamic game characterized by the DNN itself. Through this lens, different classes of optimizers can be seen as matching different types of Nash equilibria, depending on the implicit information structure of each (p)layer. The resulting method, called Dynamic Game Theoretic Neural Optimizer (DGNOpt), not only generalizes OCT-inspired optimizers to richer network class; it also motivates a new training principle by solving a multi-player cooperative game. DGNOpt shows convergence improvements over existing methods on image classification datasets with residual and inception networks. Our work marries strengths from both OCT and game theory, paving ways to new algorithmic opportunities from robust optimal control and bandit-based optimization." @default.
- W3169482895 created "2021-06-22" @default.
- W3169482895 creator A5031381142 @default.
- W3169482895 creator A5044505993 @default.
- W3169482895 creator A5087793625 @default.
- W3169482895 date "2021-07-18" @default.
- W3169482895 modified "2023-09-26" @default.
- W3169482895 title "Dynamic Game Theoretic Neural Optimizer" @default.
- W3169482895 hasPublicationYear "2021" @default.
- W3169482895 type Work @default.
- W3169482895 sameAs 3169482895 @default.
- W3169482895 citedByCount "3" @default.
- W3169482895 countsByYear W31694828952020 @default.
- W3169482895 countsByYear W31694828952021 @default.
- W3169482895 crossrefType "proceedings-article" @default.
- W3169482895 hasAuthorship W3169482895A5031381142 @default.
- W3169482895 hasAuthorship W3169482895A5044505993 @default.
- W3169482895 hasAuthorship W3169482895A5087793625 @default.
- W3169482895 hasConcept C105795698 @default.
- W3169482895 hasConcept C11413529 @default.
- W3169482895 hasConcept C126255220 @default.
- W3169482895 hasConcept C144237770 @default.
- W3169482895 hasConcept C154945302 @default.
- W3169482895 hasConcept C155032097 @default.
- W3169482895 hasConcept C155512373 @default.
- W3169482895 hasConcept C162324750 @default.
- W3169482895 hasConcept C177142836 @default.
- W3169482895 hasConcept C2777303404 @default.
- W3169482895 hasConcept C2780598303 @default.
- W3169482895 hasConcept C33923547 @default.
- W3169482895 hasConcept C41008148 @default.
- W3169482895 hasConcept C46814582 @default.
- W3169482895 hasConcept C50522688 @default.
- W3169482895 hasConcept C50644808 @default.
- W3169482895 hasConcept C73795354 @default.
- W3169482895 hasConcept C80444323 @default.
- W3169482895 hasConceptScore W3169482895C105795698 @default.
- W3169482895 hasConceptScore W3169482895C11413529 @default.
- W3169482895 hasConceptScore W3169482895C126255220 @default.
- W3169482895 hasConceptScore W3169482895C144237770 @default.
- W3169482895 hasConceptScore W3169482895C154945302 @default.
- W3169482895 hasConceptScore W3169482895C155032097 @default.
- W3169482895 hasConceptScore W3169482895C155512373 @default.
- W3169482895 hasConceptScore W3169482895C162324750 @default.
- W3169482895 hasConceptScore W3169482895C177142836 @default.
- W3169482895 hasConceptScore W3169482895C2777303404 @default.
- W3169482895 hasConceptScore W3169482895C2780598303 @default.
- W3169482895 hasConceptScore W3169482895C33923547 @default.
- W3169482895 hasConceptScore W3169482895C41008148 @default.
- W3169482895 hasConceptScore W3169482895C46814582 @default.
- W3169482895 hasConceptScore W3169482895C50522688 @default.
- W3169482895 hasConceptScore W3169482895C50644808 @default.
- W3169482895 hasConceptScore W3169482895C73795354 @default.
- W3169482895 hasConceptScore W3169482895C80444323 @default.
- W3169482895 hasLocation W31694828951 @default.
- W3169482895 hasOpenAccess W3169482895 @default.
- W3169482895 hasPrimaryLocation W31694828951 @default.
- W3169482895 hasRelatedWork W1986154481 @default.
- W3169482895 hasRelatedWork W2036575878 @default.
- W3169482895 hasRelatedWork W2095462297 @default.
- W3169482895 hasRelatedWork W2149519648 @default.
- W3169482895 hasRelatedWork W2207792509 @default.
- W3169482895 hasRelatedWork W2363632058 @default.
- W3169482895 hasRelatedWork W2462044286 @default.
- W3169482895 hasRelatedWork W2803652442 @default.
- W3169482895 hasRelatedWork W2908150527 @default.
- W3169482895 hasRelatedWork W2915829945 @default.
- W3169482895 hasRelatedWork W3007720577 @default.
- W3169482895 hasRelatedWork W3037036469 @default.
- W3169482895 hasRelatedWork W3096274457 @default.
- W3169482895 hasRelatedWork W3104068196 @default.
- W3169482895 hasRelatedWork W3115226733 @default.
- W3169482895 hasRelatedWork W3138162409 @default.
- W3169482895 hasRelatedWork W3161605098 @default.
- W3169482895 hasRelatedWork W3176887806 @default.
- W3169482895 hasRelatedWork W3184804348 @default.
- W3169482895 hasRelatedWork W64873114 @default.
- W3169482895 isParatext "false" @default.
- W3169482895 isRetracted "false" @default.
- W3169482895 magId "3169482895" @default.
- W3169482895 workType "article" @default.