Matches in SemOpenAlex for { <https://semopenalex.org/work/W3039204554> ?p ?o ?g. }
Showing items 1 to 88 of
88
with 100 items per page.
- W3039204554 endingPage "99" @default.
- W3039204554 startingPage "85" @default.
- W3039204554 abstract "The accuracy of deep learning, i.e., deep neural networks, can be characterized by dividing the total error into three main types: approximation error, optimization error, and generalization error. Whereas there are some satisfactory answers to the problems of approximation and optimization, much less is known about the theory of generalization. Most existing theoretical works for generalization fail to explain the performance of neural networks in practice. To derive a meaningful bound, we study the generalization error of neural networks for classification problems in terms of data distribution and neural network smoothness. We introduce the cover complexity (CC) to measure the difficulty of learning a data set and the inverse of the modulus of continuity to quantify neural network smoothness. A quantitative bound for expected accuracy/error is derived by considering both the CC and neural network smoothness. Although most of the analysis is general and not specific to neural networks, we validate our theoretical assumptions and results numerically for neural networks by several data sets of images. The numerical results confirm that the expected error of trained networks scaled with the square root of the number of classes has a linear relationship with respect to the CC. We also observe a clear consistency between test loss and neural network smoothness during the training process. In addition, we demonstrate empirically that the neural network smoothness decreases when the network size increases whereas the smoothness is insensitive to training dataset size." @default.
- W3039204554 created "2020-07-10" @default.
- W3039204554 creator A5000621481 @default.
- W3039204554 creator A5008536963 @default.
- W3039204554 creator A5009632487 @default.
- W3039204554 creator A5009658255 @default.
- W3039204554 date "2020-10-01" @default.
- W3039204554 modified "2023-09-23" @default.
- W3039204554 title "Quantifying the generalization error in deep learning in terms of data distribution and neural network smoothness" @default.
- W3039204554 cites W163763094 @default.
- W3039204554 cites W2103496339 @default.
- W3039204554 cites W2112796928 @default.
- W3039204554 cites W2137983211 @default.
- W3039204554 cites W2257979135 @default.
- W3039204554 cites W2529714286 @default.
- W3039204554 cites W2783538964 @default.
- W3039204554 cites W2996320484 @default.
- W3039204554 doi "https://doi.org/10.1016/j.neunet.2020.06.024" @default.
- W3039204554 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/32650153" @default.
- W3039204554 hasPublicationYear "2020" @default.
- W3039204554 type Work @default.
- W3039204554 sameAs 3039204554 @default.
- W3039204554 citedByCount "29" @default.
- W3039204554 countsByYear W30392045542019 @default.
- W3039204554 countsByYear W30392045542020 @default.
- W3039204554 countsByYear W30392045542021 @default.
- W3039204554 countsByYear W30392045542022 @default.
- W3039204554 countsByYear W30392045542023 @default.
- W3039204554 crossrefType "journal-article" @default.
- W3039204554 hasAuthorship W3039204554A5000621481 @default.
- W3039204554 hasAuthorship W3039204554A5008536963 @default.
- W3039204554 hasAuthorship W3039204554A5009632487 @default.
- W3039204554 hasAuthorship W3039204554A5009658255 @default.
- W3039204554 hasBestOaLocation W30392045541 @default.
- W3039204554 hasConcept C102634674 @default.
- W3039204554 hasConcept C105795698 @default.
- W3039204554 hasConcept C108583219 @default.
- W3039204554 hasConcept C11413529 @default.
- W3039204554 hasConcept C122383733 @default.
- W3039204554 hasConcept C134306372 @default.
- W3039204554 hasConcept C139945424 @default.
- W3039204554 hasConcept C154945302 @default.
- W3039204554 hasConcept C177148314 @default.
- W3039204554 hasConcept C28826006 @default.
- W3039204554 hasConcept C33923547 @default.
- W3039204554 hasConcept C41008148 @default.
- W3039204554 hasConcept C50644808 @default.
- W3039204554 hasConceptScore W3039204554C102634674 @default.
- W3039204554 hasConceptScore W3039204554C105795698 @default.
- W3039204554 hasConceptScore W3039204554C108583219 @default.
- W3039204554 hasConceptScore W3039204554C11413529 @default.
- W3039204554 hasConceptScore W3039204554C122383733 @default.
- W3039204554 hasConceptScore W3039204554C134306372 @default.
- W3039204554 hasConceptScore W3039204554C139945424 @default.
- W3039204554 hasConceptScore W3039204554C154945302 @default.
- W3039204554 hasConceptScore W3039204554C177148314 @default.
- W3039204554 hasConceptScore W3039204554C28826006 @default.
- W3039204554 hasConceptScore W3039204554C33923547 @default.
- W3039204554 hasConceptScore W3039204554C41008148 @default.
- W3039204554 hasConceptScore W3039204554C50644808 @default.
- W3039204554 hasFunder F4320306084 @default.
- W3039204554 hasFunder F4320321001 @default.
- W3039204554 hasFunder F4320321540 @default.
- W3039204554 hasFunder F4320337531 @default.
- W3039204554 hasFunder F4320338279 @default.
- W3039204554 hasLocation W30392045541 @default.
- W3039204554 hasLocation W30392045542 @default.
- W3039204554 hasLocation W30392045543 @default.
- W3039204554 hasLocation W30392045544 @default.
- W3039204554 hasOpenAccess W3039204554 @default.
- W3039204554 hasPrimaryLocation W30392045541 @default.
- W3039204554 hasRelatedWork W2037316683 @default.
- W3039204554 hasRelatedWork W2099878889 @default.
- W3039204554 hasRelatedWork W2188032833 @default.
- W3039204554 hasRelatedWork W2395675490 @default.
- W3039204554 hasRelatedWork W2807954395 @default.
- W3039204554 hasRelatedWork W2808471159 @default.
- W3039204554 hasRelatedWork W2971305136 @default.
- W3039204554 hasRelatedWork W3173604411 @default.
- W3039204554 hasRelatedWork W4238036087 @default.
- W3039204554 hasRelatedWork W4377969223 @default.
- W3039204554 hasVolume "130" @default.
- W3039204554 isParatext "false" @default.
- W3039204554 isRetracted "false" @default.
- W3039204554 magId "3039204554" @default.
- W3039204554 workType "article" @default.