Matches in SemOpenAlex for { <https://semopenalex.org/work/W2911892705> ?p ?o ?g. }
- W2911892705 abstract "We propose tensorial neural networks (TNNs), a generalization of existing neural networks by extending tensor operations on low order operands to those on high order ones. The problem of parameter learning is challenging, as it corresponds to hierarchical nonlinear tensor decomposition. We propose to solve the learning problem using stochastic gradient descent by deriving nontrivial backpropagation rules in generalized tensor algebra we introduce. Our proposed TNNs has three advantages over existing neural networks: (1) TNNs naturally apply to high order input object and thus preserve the multi-dimensional structure in the input, as there is no need to flatten the data. (2) TNNs interpret designs of existing neural network architectures. (3) Mapping a neural network to TNNs with the same expressive power results in a TNN of fewer parameters. TNN based compression of neural network improves existing low-rank approximation based compression methods as TNNs exploit two other types of invariant structures, periodicity and modulation, in addition to the low rankness. Experiments on LeNet-5 (MNIST), ResNet-32 (CIFAR10) and ResNet-50 (ImageNet) demonstrate that our TNN based compression outperforms (5% test accuracy improvement universally on CIFAR10) the state-of-the-art low-rank approximation based compression methods under the same compression rate, besides achieving orders of magnitude faster convergence rates due to the efficiency of TNNs." @default.
- W2911892705 created "2019-02-21" @default.
- W2911892705 creator A5008267969 @default.
- W2911892705 creator A5010098535 @default.
- W2911892705 creator A5019096316 @default.
- W2911892705 creator A5037162168 @default.
- W2911892705 date "2018-05-25" @default.
- W2911892705 modified "2023-09-27" @default.
- W2911892705 title "Tensorial Neural Networks: Generalization of Neural Networks and Application to Model Compression" @default.
- W2911892705 cites W1522301498 @default.
- W2911892705 cites W1686810756 @default.
- W2911892705 cites W1690739335 @default.
- W2911892705 cites W1799366690 @default.
- W2911892705 cites W1821462560 @default.
- W2911892705 cites W1902041153 @default.
- W2911892705 cites W1902934009 @default.
- W2911892705 cites W1955967056 @default.
- W2911892705 cites W1978376511 @default.
- W2911892705 cites W2024165284 @default.
- W2911892705 cites W2025603201 @default.
- W2911892705 cites W2112796928 @default.
- W2911892705 cites W2131524184 @default.
- W2911892705 cites W2167215970 @default.
- W2911892705 cites W2516041031 @default.
- W2911892705 cites W2559813832 @default.
- W2911892705 cites W2560017826 @default.
- W2911892705 cites W2617994470 @default.
- W2911892705 cites W2626373159 @default.
- W2911892705 cites W2766808838 @default.
- W2911892705 cites W2766839578 @default.
- W2911892705 cites W2779200694 @default.
- W2911892705 cites W2803023299 @default.
- W2911892705 cites W2949650786 @default.
- W2911892705 cites W2950967261 @default.
- W2911892705 cites W2951583185 @default.
- W2911892705 cites W2952689122 @default.
- W2911892705 cites W2953106684 @default.
- W2911892705 cites W2962988160 @default.
- W2911892705 cites W2963446712 @default.
- W2911892705 cites W2964299589 @default.
- W2911892705 cites W3101570982 @default.
- W2911892705 hasPublicationYear "2018" @default.
- W2911892705 type Work @default.
- W2911892705 sameAs 2911892705 @default.
- W2911892705 citedByCount "9" @default.
- W2911892705 countsByYear W29118927052019 @default.
- W2911892705 countsByYear W29118927052020 @default.
- W2911892705 countsByYear W29118927052021 @default.
- W2911892705 crossrefType "posted-content" @default.
- W2911892705 hasAuthorship W2911892705A5008267969 @default.
- W2911892705 hasAuthorship W2911892705A5010098535 @default.
- W2911892705 hasAuthorship W2911892705A5019096316 @default.
- W2911892705 hasAuthorship W2911892705A5037162168 @default.
- W2911892705 hasConcept C11413529 @default.
- W2911892705 hasConcept C134306372 @default.
- W2911892705 hasConcept C154945302 @default.
- W2911892705 hasConcept C155032097 @default.
- W2911892705 hasConcept C155281189 @default.
- W2911892705 hasConcept C177148314 @default.
- W2911892705 hasConcept C190502265 @default.
- W2911892705 hasConcept C202444582 @default.
- W2911892705 hasConcept C33923547 @default.
- W2911892705 hasConcept C41008148 @default.
- W2911892705 hasConcept C50644808 @default.
- W2911892705 hasConcept C78548338 @default.
- W2911892705 hasConcept C80444323 @default.
- W2911892705 hasConceptScore W2911892705C11413529 @default.
- W2911892705 hasConceptScore W2911892705C134306372 @default.
- W2911892705 hasConceptScore W2911892705C154945302 @default.
- W2911892705 hasConceptScore W2911892705C155032097 @default.
- W2911892705 hasConceptScore W2911892705C155281189 @default.
- W2911892705 hasConceptScore W2911892705C177148314 @default.
- W2911892705 hasConceptScore W2911892705C190502265 @default.
- W2911892705 hasConceptScore W2911892705C202444582 @default.
- W2911892705 hasConceptScore W2911892705C33923547 @default.
- W2911892705 hasConceptScore W2911892705C41008148 @default.
- W2911892705 hasConceptScore W2911892705C50644808 @default.
- W2911892705 hasConceptScore W2911892705C78548338 @default.
- W2911892705 hasConceptScore W2911892705C80444323 @default.
- W2911892705 hasLocation W29118927051 @default.
- W2911892705 hasOpenAccess W2911892705 @default.
- W2911892705 hasPrimaryLocation W29118927051 @default.
- W2911892705 hasRelatedWork W1798945469 @default.
- W2911892705 hasRelatedWork W1993482030 @default.
- W2911892705 hasRelatedWork W2024165284 @default.
- W2911892705 hasRelatedWork W2104636679 @default.
- W2911892705 hasRelatedWork W2112796928 @default.
- W2911892705 hasRelatedWork W2131524184 @default.
- W2911892705 hasRelatedWork W2161591461 @default.
- W2911892705 hasRelatedWork W2163605009 @default.
- W2911892705 hasRelatedWork W2167215970 @default.
- W2911892705 hasRelatedWork W2194775991 @default.
- W2911892705 hasRelatedWork W2431890537 @default.
- W2911892705 hasRelatedWork W2559813832 @default.
- W2911892705 hasRelatedWork W2803450289 @default.
- W2911892705 hasRelatedWork W2950967261 @default.
- W2911892705 hasRelatedWork W2962988160 @default.
- W2911892705 hasRelatedWork W2963363373 @default.
- W2911892705 hasRelatedWork W2963704562 @default.
- W2911892705 hasRelatedWork W2964299589 @default.