Matches in SemOpenAlex for { <https://semopenalex.org/work/W2803625257> ?p ?o ?g. }
Showing items 1 to 93 of
93
with 100 items per page.
- W2803625257 abstract "Future mobile devices are anticipated to perceive, understand and react to the world on their own by running multiple correlated deep neural networks on-device. Yet the complexity of these neural networks needs to be trimmed down both within-model and cross-model to fit in mobile storage and memory. Previous studies focus on squeezing the redundancy within a single neural network. In this work, we aim to reduce the redundancy across multiple models. We propose Multi-Task Zipping (MTZ), a framework to automatically merge correlated, pre-trained deep neural networks for cross-model compression. Central in MTZ is a layer-wise neuron sharing and incoming weight updating scheme that induces a minimal change in the error function. MTZ inherits information from each model and demands light retraining to re-boost the accuracy of individual tasks. Evaluations show that MTZ is able to fully merge the hidden layers of two VGG-16 networks with a 3.18% increase in the test error averaged on ImageNet and CelebA, or share 39.61% parameters between the two networks with <0.5% increase in the test errors for both tasks. The number of iterations to retrain the combined network is at least 17.8 times lower than that of training a single VGG-16 network. Moreover, experiments show that MTZ is also able to effectively merge multiple residual networks." @default.
- W2803625257 created "2018-06-01" @default.
- W2803625257 creator A5009841558 @default.
- W2803625257 creator A5011140675 @default.
- W2803625257 creator A5060999697 @default.
- W2803625257 date "2018-05-24" @default.
- W2803625257 modified "2023-09-23" @default.
- W2803625257 title "Multi-Task Zipping via Layer-wise Neuron Sharing" @default.
- W2803625257 cites W1686810756 @default.
- W2803625257 cites W1690739335 @default.
- W2803625257 cites W1821462560 @default.
- W2803625257 cites W1834627138 @default.
- W2803625257 cites W2067713319 @default.
- W2803625257 cites W2112796928 @default.
- W2803625257 cites W2114766824 @default.
- W2803625257 cites W2117539524 @default.
- W2803625257 cites W2125389748 @default.
- W2803625257 cites W2149933564 @default.
- W2803625257 cites W2161591461 @default.
- W2803625257 cites W2194321275 @default.
- W2803625257 cites W2194775991 @default.
- W2803625257 cites W2335728318 @default.
- W2803625257 cites W2401231614 @default.
- W2803625257 cites W2407277018 @default.
- W2803625257 cites W24089286 @default.
- W2803625257 cites W2495425901 @default.
- W2803625257 cites W2510725918 @default.
- W2803625257 cites W2549401308 @default.
- W2803625257 cites W2625157458 @default.
- W2803625257 cites W2742079690 @default.
- W2803625257 cites W2754395205 @default.
- W2803625257 cites W2913340405 @default.
- W2803625257 cites W2963114950 @default.
- W2803625257 cites W2963211188 @default.
- W2803625257 cites W2963877604 @default.
- W2803625257 cites W2963981420 @default.
- W2803625257 cites W2964299589 @default.
- W2803625257 cites W3118608800 @default.
- W2803625257 hasPublicationYear "2018" @default.
- W2803625257 type Work @default.
- W2803625257 sameAs 2803625257 @default.
- W2803625257 citedByCount "3" @default.
- W2803625257 countsByYear W28036252572019 @default.
- W2803625257 countsByYear W28036252572020 @default.
- W2803625257 crossrefType "posted-content" @default.
- W2803625257 hasAuthorship W2803625257A5009841558 @default.
- W2803625257 hasAuthorship W2803625257A5011140675 @default.
- W2803625257 hasAuthorship W2803625257A5060999697 @default.
- W2803625257 hasConcept C111919701 @default.
- W2803625257 hasConcept C11413529 @default.
- W2803625257 hasConcept C119857082 @default.
- W2803625257 hasConcept C152124472 @default.
- W2803625257 hasConcept C154945302 @default.
- W2803625257 hasConcept C155512373 @default.
- W2803625257 hasConcept C173608175 @default.
- W2803625257 hasConcept C197129107 @default.
- W2803625257 hasConcept C41008148 @default.
- W2803625257 hasConcept C50644808 @default.
- W2803625257 hasConceptScore W2803625257C111919701 @default.
- W2803625257 hasConceptScore W2803625257C11413529 @default.
- W2803625257 hasConceptScore W2803625257C119857082 @default.
- W2803625257 hasConceptScore W2803625257C152124472 @default.
- W2803625257 hasConceptScore W2803625257C154945302 @default.
- W2803625257 hasConceptScore W2803625257C155512373 @default.
- W2803625257 hasConceptScore W2803625257C173608175 @default.
- W2803625257 hasConceptScore W2803625257C197129107 @default.
- W2803625257 hasConceptScore W2803625257C41008148 @default.
- W2803625257 hasConceptScore W2803625257C50644808 @default.
- W2803625257 hasOpenAccess W2803625257 @default.
- W2803625257 hasRelatedWork W1527646298 @default.
- W2803625257 hasRelatedWork W1931611617 @default.
- W2803625257 hasRelatedWork W2160082767 @default.
- W2803625257 hasRelatedWork W2495425901 @default.
- W2803625257 hasRelatedWork W2525340990 @default.
- W2803625257 hasRelatedWork W2583761661 @default.
- W2803625257 hasRelatedWork W2768666611 @default.
- W2803625257 hasRelatedWork W2773352775 @default.
- W2803625257 hasRelatedWork W2796346584 @default.
- W2803625257 hasRelatedWork W2891874923 @default.
- W2803625257 hasRelatedWork W2921245078 @default.
- W2803625257 hasRelatedWork W2951401169 @default.
- W2803625257 hasRelatedWork W2963748143 @default.
- W2803625257 hasRelatedWork W2964024558 @default.
- W2803625257 hasRelatedWork W2965112580 @default.
- W2803625257 hasRelatedWork W2972899427 @default.
- W2803625257 hasRelatedWork W3034538858 @default.
- W2803625257 hasRelatedWork W3163875631 @default.
- W2803625257 hasRelatedWork W3180849877 @default.
- W2803625257 hasRelatedWork W2957984704 @default.
- W2803625257 isParatext "false" @default.
- W2803625257 isRetracted "false" @default.
- W2803625257 magId "2803625257" @default.
- W2803625257 workType "article" @default.