Matches in SemOpenAlex for { <https://semopenalex.org/work/W3047184523> ?p ?o ?g. }
- W3047184523 abstract "In this paper, we tackle an open research question in transfer learning, which is selecting a model initialization to achieve high performance on a new task, given several pre-trained models. We propose a new highly efficient and accurate approach based on duality diagram similarity (DDS) between deep neural networks (DNNs). DDS is a generic framework to represent and compare data of different feature dimensions. We validate our approach on the Taskonomy dataset by measuring the correspondence between actual transfer learning performance rankings on 17 taskonomy tasks and predicted rankings. Computing DDS based ranking for $17times17$ transfers requires less than 2 minutes and shows a high correlation ($0.86$) with actual transfer learning rankings, outperforming state-of-the-art methods by a large margin ($10%$) on the Taskonomy benchmark. We also demonstrate the robustness of our model selection approach to a new task, namely Pascal VOC semantic segmentation. Additionally, we show that our method can be applied to select the best layer locations within a DNN for transfer learning on 2D, 3D and semantic tasks on NYUv2 and Pascal VOC datasets." @default.
- W3047184523 created "2020-08-10" @default.
- W3047184523 creator A5024014108 @default.
- W3047184523 creator A5025034643 @default.
- W3047184523 creator A5061127696 @default.
- W3047184523 creator A5089543782 @default.
- W3047184523 date "2020-08-05" @default.
- W3047184523 modified "2023-09-27" @default.
- W3047184523 title "Duality Diagram Similarity: a generic framework for initialization selection in task transfer learning" @default.
- W3047184523 cites W125693051 @default.
- W3047184523 cites W1487583988 @default.
- W3047184523 cites W1495781036 @default.
- W3047184523 cites W1787224781 @default.
- W3047184523 cites W1836465849 @default.
- W3047184523 cites W2031489346 @default.
- W3047184523 cites W2145023731 @default.
- W3047184523 cites W2151507715 @default.
- W3047184523 cites W2165698076 @default.
- W3047184523 cites W2194775991 @default.
- W3047184523 cites W2203224402 @default.
- W3047184523 cites W2299668505 @default.
- W3047184523 cites W2312004824 @default.
- W3047184523 cites W2346578521 @default.
- W3047184523 cites W2502312327 @default.
- W3047184523 cites W2510153535 @default.
- W3047184523 cites W2531044305 @default.
- W3047184523 cites W2593768305 @default.
- W3047184523 cites W2603777577 @default.
- W3047184523 cites W2613718673 @default.
- W3047184523 cites W2618530766 @default.
- W3047184523 cites W2630837129 @default.
- W3047184523 cites W2767179670 @default.
- W3047184523 cites W2767204723 @default.
- W3047184523 cites W2942810103 @default.
- W3047184523 cites W2962835968 @default.
- W3047184523 cites W2962864421 @default.
- W3047184523 cites W2962945654 @default.
- W3047184523 cites W2963503333 @default.
- W3047184523 cites W2963591054 @default.
- W3047184523 cites W2964185501 @default.
- W3047184523 cites W2970120288 @default.
- W3047184523 cites W29779864 @default.
- W3047184523 cites W2982303846 @default.
- W3047184523 cites W2984037482 @default.
- W3047184523 cites W3105853563 @default.
- W3047184523 cites W845365781 @default.
- W3047184523 hasPublicationYear "2020" @default.
- W3047184523 type Work @default.
- W3047184523 sameAs 3047184523 @default.
- W3047184523 citedByCount "1" @default.
- W3047184523 countsByYear W30471845232020 @default.
- W3047184523 crossrefType "posted-content" @default.
- W3047184523 hasAuthorship W3047184523A5024014108 @default.
- W3047184523 hasAuthorship W3047184523A5025034643 @default.
- W3047184523 hasAuthorship W3047184523A5061127696 @default.
- W3047184523 hasAuthorship W3047184523A5089543782 @default.
- W3047184523 hasConcept C104317684 @default.
- W3047184523 hasConcept C114466953 @default.
- W3047184523 hasConcept C119857082 @default.
- W3047184523 hasConcept C13280743 @default.
- W3047184523 hasConcept C150899416 @default.
- W3047184523 hasConcept C154945302 @default.
- W3047184523 hasConcept C185592680 @default.
- W3047184523 hasConcept C185798385 @default.
- W3047184523 hasConcept C199360897 @default.
- W3047184523 hasConcept C205649164 @default.
- W3047184523 hasConcept C41008148 @default.
- W3047184523 hasConcept C55493867 @default.
- W3047184523 hasConcept C63479239 @default.
- W3047184523 hasConcept C75608658 @default.
- W3047184523 hasConcept C774472 @default.
- W3047184523 hasConceptScore W3047184523C104317684 @default.
- W3047184523 hasConceptScore W3047184523C114466953 @default.
- W3047184523 hasConceptScore W3047184523C119857082 @default.
- W3047184523 hasConceptScore W3047184523C13280743 @default.
- W3047184523 hasConceptScore W3047184523C150899416 @default.
- W3047184523 hasConceptScore W3047184523C154945302 @default.
- W3047184523 hasConceptScore W3047184523C185592680 @default.
- W3047184523 hasConceptScore W3047184523C185798385 @default.
- W3047184523 hasConceptScore W3047184523C199360897 @default.
- W3047184523 hasConceptScore W3047184523C205649164 @default.
- W3047184523 hasConceptScore W3047184523C41008148 @default.
- W3047184523 hasConceptScore W3047184523C55493867 @default.
- W3047184523 hasConceptScore W3047184523C63479239 @default.
- W3047184523 hasConceptScore W3047184523C75608658 @default.
- W3047184523 hasConceptScore W3047184523C774472 @default.
- W3047184523 hasLocation W30471845231 @default.
- W3047184523 hasOpenAccess W3047184523 @default.
- W3047184523 hasPrimaryLocation W30471845231 @default.
- W3047184523 hasRelatedWork W1554603877 @default.
- W3047184523 hasRelatedWork W1666772023 @default.
- W3047184523 hasRelatedWork W1814624917 @default.
- W3047184523 hasRelatedWork W2024922534 @default.
- W3047184523 hasRelatedWork W2121095699 @default.
- W3047184523 hasRelatedWork W2200774174 @default.
- W3047184523 hasRelatedWork W2468718006 @default.
- W3047184523 hasRelatedWork W2512569467 @default.
- W3047184523 hasRelatedWork W2579412072 @default.
- W3047184523 hasRelatedWork W2742502387 @default.
- W3047184523 hasRelatedWork W2768169601 @default.