Matches in SemOpenAlex for { <https://semopenalex.org/work/W2963497908> ?p ?o ?g. }
Showing items 1 to 84 of
84
with 100 items per page.
- W2963497908 abstract "We present a novel layerwise optimization algorithm for the learning objective of Piecewise-Linear Convolutional Neural Networks (PL-CNNs), a large class of convolutional neural networks. Specifically, PL-CNNs employ piecewise linear non-linearities such as the commonly used ReLU and max-pool, and an SVM classifier as the final layer. The key observation of our approach is that the problem corresponding to the parameter estimation of a layer can be formulated as a difference-of-convex (DC) program, which happens to be a latent structured SVM. We optimize the DC program using the concave-convex procedure, which requires us to iteratively solve a structured SVM problem. This allows to design an optimization algorithm with an optimal learning rate that does not require any tuning. Using the MNIST, CIFAR and ImageNet data sets, we show that our approach always improves over the state of the art variants of backpropagation and scales to large data and large network settings." @default.
- W2963497908 created "2019-07-30" @default.
- W2963497908 creator A5018728658 @default.
- W2963497908 creator A5049231060 @default.
- W2963497908 creator A5057678172 @default.
- W2963497908 date "2016-11-04" @default.
- W2963497908 modified "2023-09-27" @default.
- W2963497908 title "Trusting SVM for Piecewise Linear CNNs" @default.
- W2963497908 hasPublicationYear "2016" @default.
- W2963497908 type Work @default.
- W2963497908 sameAs 2963497908 @default.
- W2963497908 citedByCount "1" @default.
- W2963497908 countsByYear W29634979082018 @default.
- W2963497908 crossrefType "proceedings-article" @default.
- W2963497908 hasAuthorship W2963497908A5018728658 @default.
- W2963497908 hasAuthorship W2963497908A5049231060 @default.
- W2963497908 hasAuthorship W2963497908A5057678172 @default.
- W2963497908 hasConcept C112680207 @default.
- W2963497908 hasConcept C11413529 @default.
- W2963497908 hasConcept C119857082 @default.
- W2963497908 hasConcept C12267149 @default.
- W2963497908 hasConcept C126255220 @default.
- W2963497908 hasConcept C134306372 @default.
- W2963497908 hasConcept C145446738 @default.
- W2963497908 hasConcept C153180895 @default.
- W2963497908 hasConcept C154945302 @default.
- W2963497908 hasConcept C155032097 @default.
- W2963497908 hasConcept C157972887 @default.
- W2963497908 hasConcept C164660894 @default.
- W2963497908 hasConcept C17095337 @default.
- W2963497908 hasConcept C190502265 @default.
- W2963497908 hasConcept C2524010 @default.
- W2963497908 hasConcept C33923547 @default.
- W2963497908 hasConcept C41008148 @default.
- W2963497908 hasConcept C50644808 @default.
- W2963497908 hasConcept C81363708 @default.
- W2963497908 hasConcept C95623464 @default.
- W2963497908 hasConceptScore W2963497908C112680207 @default.
- W2963497908 hasConceptScore W2963497908C11413529 @default.
- W2963497908 hasConceptScore W2963497908C119857082 @default.
- W2963497908 hasConceptScore W2963497908C12267149 @default.
- W2963497908 hasConceptScore W2963497908C126255220 @default.
- W2963497908 hasConceptScore W2963497908C134306372 @default.
- W2963497908 hasConceptScore W2963497908C145446738 @default.
- W2963497908 hasConceptScore W2963497908C153180895 @default.
- W2963497908 hasConceptScore W2963497908C154945302 @default.
- W2963497908 hasConceptScore W2963497908C155032097 @default.
- W2963497908 hasConceptScore W2963497908C157972887 @default.
- W2963497908 hasConceptScore W2963497908C164660894 @default.
- W2963497908 hasConceptScore W2963497908C17095337 @default.
- W2963497908 hasConceptScore W2963497908C190502265 @default.
- W2963497908 hasConceptScore W2963497908C2524010 @default.
- W2963497908 hasConceptScore W2963497908C33923547 @default.
- W2963497908 hasConceptScore W2963497908C41008148 @default.
- W2963497908 hasConceptScore W2963497908C50644808 @default.
- W2963497908 hasConceptScore W2963497908C81363708 @default.
- W2963497908 hasConceptScore W2963497908C95623464 @default.
- W2963497908 hasLocation W29634979081 @default.
- W2963497908 hasOpenAccess W2963497908 @default.
- W2963497908 hasPrimaryLocation W29634979081 @default.
- W2963497908 hasRelatedWork W152607948 @default.
- W2963497908 hasRelatedWork W1991094229 @default.
- W2963497908 hasRelatedWork W1999910732 @default.
- W2963497908 hasRelatedWork W2003124853 @default.
- W2963497908 hasRelatedWork W2096114681 @default.
- W2963497908 hasRelatedWork W2101577816 @default.
- W2963497908 hasRelatedWork W2112353934 @default.
- W2963497908 hasRelatedWork W2141355352 @default.
- W2963497908 hasRelatedWork W2188977358 @default.
- W2963497908 hasRelatedWork W2189508540 @default.
- W2963497908 hasRelatedWork W2291079862 @default.
- W2963497908 hasRelatedWork W2554683425 @default.
- W2963497908 hasRelatedWork W2794311605 @default.
- W2963497908 hasRelatedWork W2806448755 @default.
- W2963497908 hasRelatedWork W3006720434 @default.
- W2963497908 hasRelatedWork W3035715279 @default.
- W2963497908 hasRelatedWork W3132376261 @default.
- W2963497908 hasRelatedWork W3176726106 @default.
- W2963497908 hasRelatedWork W78648781 @default.
- W2963497908 hasRelatedWork W2983127909 @default.
- W2963497908 isParatext "false" @default.
- W2963497908 isRetracted "false" @default.
- W2963497908 magId "2963497908" @default.
- W2963497908 workType "article" @default.