FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2788351666> ?p ?o ?g. }

• W2788351666 abstract "We propose to optimize the activation functions of a deep neural network by adding a corresponding functional regularization to the cost function. We justify the use of a second-order total-variation criterion. This allows us to derive a general representer theorem for deep neural networks that makes a direct connection with splines and sparsity. Specifically, we show that the optimal network configuration can be achieved with activation functions that are nonuniform linear splines with adaptive knots. The bottom line is that the action of each neuron is encoded by a spline whose parameters (including the number of knots) are optimized during the training procedure. The scheme results in a computational structure that is compatible with the existing deep-ReLU, parametric ReLU, APL (adaptive piecewise-linear) and MaxOut architectures. It also suggests novel optimization challenges, while making the link with $ell_1$ minimization and sparsity-promoting techniques explicit." @default.
• W2788351666 created "2018-03-06" @default.
• W2788351666 creator A5027830035 @default.
• W2788351666 date "2018-02-26" @default.
• W2788351666 modified "2023-09-27" @default.
• W2788351666 title "A representer theorem for deep neural networks" @default.
• W2788351666 cites W1486164486 @default.
• W2788351666 cites W1498436455 @default.
• W2788351666 cites W1540155273 @default.
• W2788351666 cites W1560724230 @default.
• W2788351666 cites W1594563152 @default.
• W2788351666 cites W1683511521 @default.
• W2788351666 cites W1901129140 @default.
• W2788351666 cites W1986280275 @default.
• W2788351666 cites W2039402388 @default.
• W2788351666 cites W2050834445 @default.
• W2788351666 cites W2055522016 @default.
• W2788351666 cites W2065015198 @default.
• W2788351666 cites W2076063813 @default.
• W2788351666 cites W2084709495 @default.
• W2788351666 cites W2085877024 @default.
• W2788351666 cites W2086472796 @default.
• W2788351666 cites W2110652811 @default.
• W2788351666 cites W2146611938 @default.
• W2788351666 cites W2149593872 @default.
• W2788351666 cites W2156909104 @default.
• W2788351666 cites W2160815625 @default.
• W2788351666 cites W2161388792 @default.
• W2788351666 cites W2162870748 @default.
• W2788351666 cites W2163605009 @default.
• W2788351666 cites W2535756153 @default.
• W2788351666 cites W2555745781 @default.
• W2788351666 cites W2760746183 @default.
• W2788351666 cites W2786494993 @default.
• W2788351666 cites W3091064178 @default.
• W2788351666 cites W3103915668 @default.
• W2788351666 cites W39602628 @default.
• W2788351666 hasPublicationYear "2018" @default.
• W2788351666 type Work @default.
• W2788351666 sameAs 2788351666 @default.
• W2788351666 citedByCount "4" @default.
• W2788351666 countsByYear W27883516662018 @default.
• W2788351666 countsByYear W27883516662021 @default.
• W2788351666 crossrefType "posted-content" @default.
• W2788351666 hasAuthorship W2788351666A5027830035 @default.
• W2788351666 hasConcept C10390562 @default.
• W2788351666 hasConcept C105795698 @default.
• W2788351666 hasConcept C108583219 @default.
• W2788351666 hasConcept C11413529 @default.
• W2788351666 hasConcept C117251300 @default.
• W2788351666 hasConcept C122280245 @default.
• W2788351666 hasConcept C12267149 @default.
• W2788351666 hasConcept C126255220 @default.
• W2788351666 hasConcept C127413603 @default.
• W2788351666 hasConcept C134306372 @default.
• W2788351666 hasConcept C147764199 @default.
• W2788351666 hasConcept C154945302 @default.
• W2788351666 hasConcept C17095337 @default.
• W2788351666 hasConcept C172623408 @default.
• W2788351666 hasConcept C182335926 @default.
• W2788351666 hasConcept C2776135515 @default.
• W2788351666 hasConcept C2984842247 @default.
• W2788351666 hasConcept C33923547 @default.
• W2788351666 hasConcept C38365724 @default.
• W2788351666 hasConcept C41008148 @default.
• W2788351666 hasConcept C50644808 @default.
• W2788351666 hasConcept C66938386 @default.
• W2788351666 hasConceptScore W2788351666C10390562 @default.
• W2788351666 hasConceptScore W2788351666C105795698 @default.
• W2788351666 hasConceptScore W2788351666C108583219 @default.
• W2788351666 hasConceptScore W2788351666C11413529 @default.
• W2788351666 hasConceptScore W2788351666C117251300 @default.
• W2788351666 hasConceptScore W2788351666C122280245 @default.
• W2788351666 hasConceptScore W2788351666C12267149 @default.
• W2788351666 hasConceptScore W2788351666C126255220 @default.
• W2788351666 hasConceptScore W2788351666C127413603 @default.
• W2788351666 hasConceptScore W2788351666C134306372 @default.
• W2788351666 hasConceptScore W2788351666C147764199 @default.
• W2788351666 hasConceptScore W2788351666C154945302 @default.
• W2788351666 hasConceptScore W2788351666C17095337 @default.
• W2788351666 hasConceptScore W2788351666C172623408 @default.
• W2788351666 hasConceptScore W2788351666C182335926 @default.
• W2788351666 hasConceptScore W2788351666C2776135515 @default.
• W2788351666 hasConceptScore W2788351666C2984842247 @default.
• W2788351666 hasConceptScore W2788351666C33923547 @default.
• W2788351666 hasConceptScore W2788351666C38365724 @default.
• W2788351666 hasConceptScore W2788351666C41008148 @default.
• W2788351666 hasConceptScore W2788351666C50644808 @default.
• W2788351666 hasConceptScore W2788351666C66938386 @default.
• W2788351666 hasLocation W27883516661 @default.
• W2788351666 hasOpenAccess W2788351666 @default.
• W2788351666 hasPrimaryLocation W27883516661 @default.
• W2788351666 hasRelatedWork W2611375647 @default.
• W2788351666 hasRelatedWork W2749030024 @default.
• W2788351666 hasRelatedWork W2787442904 @default.
• W2788351666 hasRelatedWork W2899748887 @default.
• W2788351666 hasRelatedWork W2919767206 @default.
• W2788351666 hasRelatedWork W2963455631 @default.
• W2788351666 hasRelatedWork W2964055987 @default.
• W2788351666 hasRelatedWork W3014013215 @default.

• next