Matches in SemOpenAlex for { <https://semopenalex.org/work/W4367665094> ?p ?o ?g. }
Showing items 1 to 75 of
75
with 100 items per page.
- W4367665094 endingPage "183" @default.
- W4367665094 startingPage "171" @default.
- W4367665094 abstract "Convolutional Neural Network (CNN) models often comprise multiple layers varying in compute requirements. For deployment, a number of hardware accelerators are available that have subtle differences in compute architectures within the same family of platforms. A component (a set of layers) of a CNN model may perceive different performance in different compute architectures. Optimal mapping of the components of a CNN model across a given heterogeneous architecture can leverage underlying different compute architectures to deliver minimum inference latency $$^3$$ . In this paper, we present an optimal partitioning approach to map a CNN model across heterogeneous architectures by leveraging a repository of performance-measurement-benchmark (PerfLib) for different accelerators. Our proposed framework, Hetero-vis, decides optimal partitions and mapping of a CNN network across different accelerators to minimize the inference latency. Our experiments reveal up to 1.43 $$times $$ better performance with grouped layer deployment of CNN models on heterogeneous hardware compared to the entire model deployed on a single accelerator. $$^1$$ Inference latency and latency terms are used interchangeably in the rest of the paper." @default.
- W4367665094 created "2023-05-03" @default.
- W4367665094 creator A5040769022 @default.
- W4367665094 creator A5059412437 @default.
- W4367665094 creator A5084952766 @default.
- W4367665094 creator A5086400313 @default.
- W4367665094 date "2023-01-01" @default.
- W4367665094 modified "2023-09-26" @default.
- W4367665094 title "Hetero-Vis: A Framework for Latency Optimized Heterogeneous Deployment of Convolutional Neural Networks" @default.
- W4367665094 cites W2194775991 @default.
- W4367665094 cites W2963918968 @default.
- W4367665094 cites W2964862330 @default.
- W4367665094 cites W2967615747 @default.
- W4367665094 cites W2967733054 @default.
- W4367665094 cites W3165698711 @default.
- W4367665094 cites W3195610113 @default.
- W4367665094 doi "https://doi.org/10.1007/978-3-031-31209-0_13" @default.
- W4367665094 hasPublicationYear "2023" @default.
- W4367665094 type Work @default.
- W4367665094 citedByCount "0" @default.
- W4367665094 crossrefType "book-chapter" @default.
- W4367665094 hasAuthorship W4367665094A5040769022 @default.
- W4367665094 hasAuthorship W4367665094A5059412437 @default.
- W4367665094 hasAuthorship W4367665094A5084952766 @default.
- W4367665094 hasAuthorship W4367665094A5086400313 @default.
- W4367665094 hasConcept C105339364 @default.
- W4367665094 hasConcept C111919701 @default.
- W4367665094 hasConcept C113775141 @default.
- W4367665094 hasConcept C118524514 @default.
- W4367665094 hasConcept C120314980 @default.
- W4367665094 hasConcept C13280743 @default.
- W4367665094 hasConcept C153083717 @default.
- W4367665094 hasConcept C154945302 @default.
- W4367665094 hasConcept C173608175 @default.
- W4367665094 hasConcept C185798385 @default.
- W4367665094 hasConcept C205649164 @default.
- W4367665094 hasConcept C2776214188 @default.
- W4367665094 hasConcept C41008148 @default.
- W4367665094 hasConcept C76155785 @default.
- W4367665094 hasConcept C81363708 @default.
- W4367665094 hasConcept C82876162 @default.
- W4367665094 hasConceptScore W4367665094C105339364 @default.
- W4367665094 hasConceptScore W4367665094C111919701 @default.
- W4367665094 hasConceptScore W4367665094C113775141 @default.
- W4367665094 hasConceptScore W4367665094C118524514 @default.
- W4367665094 hasConceptScore W4367665094C120314980 @default.
- W4367665094 hasConceptScore W4367665094C13280743 @default.
- W4367665094 hasConceptScore W4367665094C153083717 @default.
- W4367665094 hasConceptScore W4367665094C154945302 @default.
- W4367665094 hasConceptScore W4367665094C173608175 @default.
- W4367665094 hasConceptScore W4367665094C185798385 @default.
- W4367665094 hasConceptScore W4367665094C205649164 @default.
- W4367665094 hasConceptScore W4367665094C2776214188 @default.
- W4367665094 hasConceptScore W4367665094C41008148 @default.
- W4367665094 hasConceptScore W4367665094C76155785 @default.
- W4367665094 hasConceptScore W4367665094C81363708 @default.
- W4367665094 hasConceptScore W4367665094C82876162 @default.
- W4367665094 hasLocation W43676650941 @default.
- W4367665094 hasOpenAccess W4367665094 @default.
- W4367665094 hasPrimaryLocation W43676650941 @default.
- W4367665094 hasRelatedWork W1505529854 @default.
- W4367665094 hasRelatedWork W1513409726 @default.
- W4367665094 hasRelatedWork W1971506073 @default.
- W4367665094 hasRelatedWork W2093683727 @default.
- W4367665094 hasRelatedWork W2139514495 @default.
- W4367665094 hasRelatedWork W2298102683 @default.
- W4367665094 hasRelatedWork W2391655055 @default.
- W4367665094 hasRelatedWork W3111137954 @default.
- W4367665094 hasRelatedWork W4291824970 @default.
- W4367665094 hasRelatedWork W4322776108 @default.
- W4367665094 isParatext "false" @default.
- W4367665094 isRetracted "false" @default.
- W4367665094 workType "book-chapter" @default.