Matches in SemOpenAlex for { <https://semopenalex.org/work/W3000593479> ?p ?o ?g. }
- W3000593479 abstract "Recently, several studies have proven the global convergence and generalization abilities of the gradient descent method for two-layer ReLU networks. Most studies especially focused on the regression problems with the squared loss function, except for a few, and the importance of the positivity of the neural tangent kernel has been pointed out. On the other hand, the performance of gradient descent on classification problems using the logistic loss function has not been well studied, and further investigation of this problem structure is possible. In this work, we demonstrate that the separability assumption using a neural tangent model is more reasonable than the positivity condition of the neural tangent kernel and provide a refined convergence analysis of the gradient descent for two-layer networks with smooth activations. A remarkable point of our result is that our convergence and generalization bounds have much better dependence on the network width in comparison to related studies. Consequently, our theory provides a generalization guarantee for less over-parameterized two-layer networks, while most studies require much higher over-parameterization." @default.
- W3000593479 created "2020-01-23" @default.
- W3000593479 creator A5023953123 @default.
- W3000593479 creator A5055409732 @default.
- W3000593479 creator A5078812767 @default.
- W3000593479 date "2022-11-28" @default.
- W3000593479 modified "2023-09-30" @default.
- W3000593479 title "Gradient Descent can Learn Less Over-parameterized Two-layer Neural Networks on Classification Problems" @default.
- W3000593479 cites W1608733719 @default.
- W3000593479 cites W1678356000 @default.
- W3000593479 cites W1869625623 @default.
- W3000593479 cites W1944672 @default.
- W3000593479 cites W1988115241 @default.
- W3000593479 cites W1996652862 @default.
- W3000593479 cites W2033855314 @default.
- W3000593479 cites W2108263314 @default.
- W3000593479 cites W2150621701 @default.
- W3000593479 cites W2591714514 @default.
- W3000593479 cites W2767286248 @default.
- W3000593479 cites W2771061327 @default.
- W3000593479 cites W2804589149 @default.
- W3000593479 cites W2809090039 @default.
- W3000593479 cites W2886067286 @default.
- W3000593479 cites W2899748887 @default.
- W3000593479 cites W2900103278 @default.
- W3000593479 cites W2900959181 @default.
- W3000593479 cites W2904838594 @default.
- W3000593479 cites W2911867426 @default.
- W3000593479 cites W2913892099 @default.
- W3000593479 cites W2914811257 @default.
- W3000593479 cites W2917744435 @default.
- W3000593479 cites W2927724204 @default.
- W3000593479 cites W2945554113 @default.
- W3000593479 cites W2946840143 @default.
- W3000593479 cites W2950987997 @default.
- W3000593479 cites W2952132225 @default.
- W3000593479 cites W2952817981 @default.
- W3000593479 cites W2954816895 @default.
- W3000593479 cites W2962767131 @default.
- W3000593479 cites W2963095610 @default.
- W3000593479 cites W2963100491 @default.
- W3000593479 cites W2963119361 @default.
- W3000593479 cites W2963623651 @default.
- W3000593479 cites W2963664410 @default.
- W3000593479 cites W2963827833 @default.
- W3000593479 cites W2964161337 @default.
- W3000593479 cites W2970032917 @default.
- W3000593479 cites W2991290085 @default.
- W3000593479 cites W2996168800 @default.
- W3000593479 cites W3101806332 @default.
- W3000593479 cites W607505555 @default.
- W3000593479 cites W2991401328 @default.
- W3000593479 hasPublicationYear "2022" @default.
- W3000593479 type Work @default.
- W3000593479 sameAs 3000593479 @default.
- W3000593479 citedByCount "18" @default.
- W3000593479 countsByYear W30005934792019 @default.
- W3000593479 countsByYear W30005934792020 @default.
- W3000593479 countsByYear W30005934792021 @default.
- W3000593479 crossrefType "posted-content" @default.
- W3000593479 hasAuthorship W3000593479A5023953123 @default.
- W3000593479 hasAuthorship W3000593479A5055409732 @default.
- W3000593479 hasAuthorship W3000593479A5078812767 @default.
- W3000593479 hasBestOaLocation W30005934791 @default.
- W3000593479 hasConcept C11413529 @default.
- W3000593479 hasConcept C119857082 @default.
- W3000593479 hasConcept C153258448 @default.
- W3000593479 hasConcept C153294291 @default.
- W3000593479 hasConcept C154945302 @default.
- W3000593479 hasConcept C165464430 @default.
- W3000593479 hasConcept C171250308 @default.
- W3000593479 hasConcept C192562407 @default.
- W3000593479 hasConcept C205649164 @default.
- W3000593479 hasConcept C2776637919 @default.
- W3000593479 hasConcept C2779227376 @default.
- W3000593479 hasConcept C41008148 @default.
- W3000593479 hasConcept C50644808 @default.
- W3000593479 hasConceptScore W3000593479C11413529 @default.
- W3000593479 hasConceptScore W3000593479C119857082 @default.
- W3000593479 hasConceptScore W3000593479C153258448 @default.
- W3000593479 hasConceptScore W3000593479C153294291 @default.
- W3000593479 hasConceptScore W3000593479C154945302 @default.
- W3000593479 hasConceptScore W3000593479C165464430 @default.
- W3000593479 hasConceptScore W3000593479C171250308 @default.
- W3000593479 hasConceptScore W3000593479C192562407 @default.
- W3000593479 hasConceptScore W3000593479C205649164 @default.
- W3000593479 hasConceptScore W3000593479C2776637919 @default.
- W3000593479 hasConceptScore W3000593479C2779227376 @default.
- W3000593479 hasConceptScore W3000593479C41008148 @default.
- W3000593479 hasConceptScore W3000593479C50644808 @default.
- W3000593479 hasLocation W30005934791 @default.
- W3000593479 hasLocation W30005934792 @default.
- W3000593479 hasOpenAccess W3000593479 @default.
- W3000593479 hasPrimaryLocation W30005934791 @default.
- W3000593479 hasRelatedWork W2040945547 @default.
- W3000593479 hasRelatedWork W2367760571 @default.
- W3000593479 hasRelatedWork W2369708740 @default.
- W3000593479 hasRelatedWork W2804589149 @default.
- W3000593479 hasRelatedWork W2952469083 @default.
- W3000593479 hasRelatedWork W3000593479 @default.