Matches in SemOpenAlex for { <https://semopenalex.org/work/W4287689794> ?p ?o ?g. }
Showing items 1 to 90 of
90
with 100 items per page.
- W4287689794 abstract "Most state of the art deep neural networks are overparameterized and exhibit a high computational cost. A straightforward approach to this problem is to replace convolutional kernels with its low-rank tensor approximations, whereas the Canonical Polyadic tensor Decomposition is one of the most suited models. However, fitting the convolutional tensors by numerical optimization algorithms often encounters diverging components, i.e., extremely large rank-one tensors but canceling each other. Such degeneracy often causes the non-interpretable result and numerical instability for the neural network fine-tuning. This paper is the first study on degeneracy in the tensor decomposition of convolutional kernels. We present a novel method, which can stabilize the low-rank approximation of convolutional kernels and ensure efficient compression while preserving the high-quality performance of the neural networks. We evaluate our approach on popular CNN architectures for image classification and show that our method results in much lower accuracy degradation and provides consistent performance." @default.
- W4287689794 created "2022-07-26" @default.
- W4287689794 creator A5004111307 @default.
- W4287689794 creator A5018676117 @default.
- W4287689794 creator A5023534854 @default.
- W4287689794 creator A5034056852 @default.
- W4287689794 creator A5045231532 @default.
- W4287689794 creator A5049652955 @default.
- W4287689794 creator A5059288464 @default.
- W4287689794 creator A5076229848 @default.
- W4287689794 creator A5086630288 @default.
- W4287689794 date "2020-08-12" @default.
- W4287689794 modified "2023-10-17" @default.
- W4287689794 title "Stable Low-rank Tensor Decomposition for Compression of Convolutional Neural Network" @default.
- W4287689794 doi "https://doi.org/10.48550/arxiv.2008.05441" @default.
- W4287689794 hasPublicationYear "2020" @default.
- W4287689794 type Work @default.
- W4287689794 citedByCount "1" @default.
- W4287689794 countsByYear W42876897942023 @default.
- W4287689794 crossrefType "posted-content" @default.
- W4287689794 hasAuthorship W4287689794A5004111307 @default.
- W4287689794 hasAuthorship W4287689794A5018676117 @default.
- W4287689794 hasAuthorship W4287689794A5023534854 @default.
- W4287689794 hasAuthorship W4287689794A5034056852 @default.
- W4287689794 hasAuthorship W4287689794A5045231532 @default.
- W4287689794 hasAuthorship W4287689794A5049652955 @default.
- W4287689794 hasAuthorship W4287689794A5059288464 @default.
- W4287689794 hasAuthorship W4287689794A5076229848 @default.
- W4287689794 hasAuthorship W4287689794A5086630288 @default.
- W4287689794 hasBestOaLocation W42876897941 @default.
- W4287689794 hasConcept C11413529 @default.
- W4287689794 hasConcept C114614502 @default.
- W4287689794 hasConcept C121332964 @default.
- W4287689794 hasConcept C124681953 @default.
- W4287689794 hasConcept C126255220 @default.
- W4287689794 hasConcept C153180895 @default.
- W4287689794 hasConcept C154945302 @default.
- W4287689794 hasConcept C155281189 @default.
- W4287689794 hasConcept C164226766 @default.
- W4287689794 hasConcept C180016635 @default.
- W4287689794 hasConcept C18903297 @default.
- W4287689794 hasConcept C202444582 @default.
- W4287689794 hasConcept C2777727622 @default.
- W4287689794 hasConcept C28826006 @default.
- W4287689794 hasConcept C2986737658 @default.
- W4287689794 hasConcept C33923547 @default.
- W4287689794 hasConcept C41008148 @default.
- W4287689794 hasConcept C42704193 @default.
- W4287689794 hasConcept C60644358 @default.
- W4287689794 hasConcept C81363708 @default.
- W4287689794 hasConcept C86803240 @default.
- W4287689794 hasConcept C97355855 @default.
- W4287689794 hasConceptScore W4287689794C11413529 @default.
- W4287689794 hasConceptScore W4287689794C114614502 @default.
- W4287689794 hasConceptScore W4287689794C121332964 @default.
- W4287689794 hasConceptScore W4287689794C124681953 @default.
- W4287689794 hasConceptScore W4287689794C126255220 @default.
- W4287689794 hasConceptScore W4287689794C153180895 @default.
- W4287689794 hasConceptScore W4287689794C154945302 @default.
- W4287689794 hasConceptScore W4287689794C155281189 @default.
- W4287689794 hasConceptScore W4287689794C164226766 @default.
- W4287689794 hasConceptScore W4287689794C180016635 @default.
- W4287689794 hasConceptScore W4287689794C18903297 @default.
- W4287689794 hasConceptScore W4287689794C202444582 @default.
- W4287689794 hasConceptScore W4287689794C2777727622 @default.
- W4287689794 hasConceptScore W4287689794C28826006 @default.
- W4287689794 hasConceptScore W4287689794C2986737658 @default.
- W4287689794 hasConceptScore W4287689794C33923547 @default.
- W4287689794 hasConceptScore W4287689794C41008148 @default.
- W4287689794 hasConceptScore W4287689794C42704193 @default.
- W4287689794 hasConceptScore W4287689794C60644358 @default.
- W4287689794 hasConceptScore W4287689794C81363708 @default.
- W4287689794 hasConceptScore W4287689794C86803240 @default.
- W4287689794 hasConceptScore W4287689794C97355855 @default.
- W4287689794 hasLocation W42876897941 @default.
- W4287689794 hasOpenAccess W4287689794 @default.
- W4287689794 hasPrimaryLocation W42876897941 @default.
- W4287689794 hasRelatedWork W2393920986 @default.
- W4287689794 hasRelatedWork W2809075421 @default.
- W4287689794 hasRelatedWork W2891735857 @default.
- W4287689794 hasRelatedWork W3048878373 @default.
- W4287689794 hasRelatedWork W3159571796 @default.
- W4287689794 hasRelatedWork W4200201902 @default.
- W4287689794 hasRelatedWork W4205920284 @default.
- W4287689794 hasRelatedWork W4283821751 @default.
- W4287689794 hasRelatedWork W4287689794 @default.
- W4287689794 hasRelatedWork W4293328203 @default.
- W4287689794 isParatext "false" @default.
- W4287689794 isRetracted "false" @default.
- W4287689794 workType "article" @default.