Matches in SemOpenAlex for { <https://semopenalex.org/work/W3206552917> ?p ?o ?g. }
- W3206552917 abstract "Neural Architecture Search (NAS) has been pivotal in finding optimal network configurations for Convolution Neural Networks (CNNs). While many methods explore NAS from a global search-space perspective, the employed optimization schemes typically require heavy computational resources. This work introduces a method that is efficient in computationally constrained environments by examining the micro-search space of channel size. In tackling channel-size optimization, we design an automated algorithm to extract the dependencies within different connected layers of the network. In addition, we introduce the idea of knowledge distillation, which enables preservation of trained weights, admist trials where the channel sizes are changing. Further, since the standard performance indicators (accuracy, loss) fail to capture the performance of individual network components (providing an overall network evaluation), we introduce a novel metric that highly correlates with test accuracy and enables analysis of individual network layers. Combining dependency extraction, metrics, and knowledge distillation, we introduce an efficient searching algorithm, with simulated annealing inspired stochasticity, and demonstrate its effectiveness in finding optimal architectures that outperform baselines by a large margin." @default.
- W3206552917 created "2021-10-25" @default.
- W3206552917 creator A5005629638 @default.
- W3206552917 creator A5008024267 @default.
- W3206552917 creator A5036842123 @default.
- W3206552917 creator A5059152392 @default.
- W3206552917 creator A5073426758 @default.
- W3206552917 date "2021-10-13" @default.
- W3206552917 modified "2023-09-27" @default.
- W3206552917 title "CONetV2: Efficient Auto-Channel Size Optimization for CNNs." @default.
- W3206552917 cites W1821462560 @default.
- W3206552917 cites W1994616650 @default.
- W3206552917 cites W2163605009 @default.
- W3206552917 cites W2333796428 @default.
- W3206552917 cites W2558580397 @default.
- W3206552917 cites W2593744649 @default.
- W3206552917 cites W2769653148 @default.
- W3206552917 cites W2773706593 @default.
- W3206552917 cites W2782417188 @default.
- W3206552917 cites W2785366763 @default.
- W3206552917 cites W2796265726 @default.
- W3206552917 cites W2803311163 @default.
- W3206552917 cites W2808938483 @default.
- W3206552917 cites W2810075754 @default.
- W3206552917 cites W2885820039 @default.
- W3206552917 cites W2899457449 @default.
- W3206552917 cites W2943530662 @default.
- W3206552917 cites W2946948417 @default.
- W3206552917 cites W2949650786 @default.
- W3206552917 cites W2949829435 @default.
- W3206552917 cites W2950179405 @default.
- W3206552917 cites W2951886768 @default.
- W3206552917 cites W2952205488 @default.
- W3206552917 cites W2963457350 @default.
- W3206552917 cites W2963628712 @default.
- W3206552917 cites W2963821229 @default.
- W3206552917 cites W2964024268 @default.
- W3206552917 cites W2964081807 @default.
- W3206552917 cites W2964118293 @default.
- W3206552917 cites W2965658867 @default.
- W3206552917 cites W2970792251 @default.
- W3206552917 cites W2982157312 @default.
- W3206552917 cites W2992043355 @default.
- W3206552917 cites W2995179471 @default.
- W3206552917 cites W3035138962 @default.
- W3206552917 cites W3035359320 @default.
- W3206552917 cites W3048844250 @default.
- W3206552917 cites W3080991741 @default.
- W3206552917 cites W3082154327 @default.
- W3206552917 cites W3083257207 @default.
- W3206552917 cites W3099343340 @default.
- W3206552917 cites W3103089845 @default.
- W3206552917 cites W3118955349 @default.
- W3206552917 cites W3165285531 @default.
- W3206552917 cites W3165945666 @default.
- W3206552917 hasPublicationYear "2021" @default.
- W3206552917 type Work @default.
- W3206552917 sameAs 3206552917 @default.
- W3206552917 citedByCount "0" @default.
- W3206552917 crossrefType "posted-content" @default.
- W3206552917 hasAuthorship W3206552917A5005629638 @default.
- W3206552917 hasAuthorship W3206552917A5008024267 @default.
- W3206552917 hasAuthorship W3206552917A5036842123 @default.
- W3206552917 hasAuthorship W3206552917A5059152392 @default.
- W3206552917 hasAuthorship W3206552917A5073426758 @default.
- W3206552917 hasConcept C119857082 @default.
- W3206552917 hasConcept C124101348 @default.
- W3206552917 hasConcept C126980161 @default.
- W3206552917 hasConcept C127162648 @default.
- W3206552917 hasConcept C127413603 @default.
- W3206552917 hasConcept C154945302 @default.
- W3206552917 hasConcept C176217482 @default.
- W3206552917 hasConcept C21547014 @default.
- W3206552917 hasConcept C31258907 @default.
- W3206552917 hasConcept C41008148 @default.
- W3206552917 hasConcept C45347329 @default.
- W3206552917 hasConcept C50644808 @default.
- W3206552917 hasConcept C774472 @default.
- W3206552917 hasConceptScore W3206552917C119857082 @default.
- W3206552917 hasConceptScore W3206552917C124101348 @default.
- W3206552917 hasConceptScore W3206552917C126980161 @default.
- W3206552917 hasConceptScore W3206552917C127162648 @default.
- W3206552917 hasConceptScore W3206552917C127413603 @default.
- W3206552917 hasConceptScore W3206552917C154945302 @default.
- W3206552917 hasConceptScore W3206552917C176217482 @default.
- W3206552917 hasConceptScore W3206552917C21547014 @default.
- W3206552917 hasConceptScore W3206552917C31258907 @default.
- W3206552917 hasConceptScore W3206552917C41008148 @default.
- W3206552917 hasConceptScore W3206552917C45347329 @default.
- W3206552917 hasConceptScore W3206552917C50644808 @default.
- W3206552917 hasConceptScore W3206552917C774472 @default.
- W3206552917 hasLocation W32065529171 @default.
- W3206552917 hasOpenAccess W3206552917 @default.
- W3206552917 hasPrimaryLocation W32065529171 @default.
- W3206552917 hasRelatedWork W1508860170 @default.
- W3206552917 hasRelatedWork W1798702550 @default.
- W3206552917 hasRelatedWork W1970755911 @default.
- W3206552917 hasRelatedWork W2061716650 @default.
- W3206552917 hasRelatedWork W2534427316 @default.
- W3206552917 hasRelatedWork W2798525957 @default.