Matches in SemOpenAlex for { <https://semopenalex.org/work/W4313288463> ?p ?o ?g. }
Showing items 1 to 80 of
80
with 100 items per page.
- W4313288463 endingPage "327" @default.
- W4313288463 startingPage "314" @default.
- W4313288463 abstract "Deep learning applications have been widely adopted on edge devices, to mitigate the privacy and latency issues of accessing cloud servers. Deciding the number of neurons during the design of a deep neural network to maximize performance is not intuitive. Particularly, many application scenarios are real-time and have a strict latency constraint, while conventional neural network optimization methods do not directly change the temporal cost of model inference for latency-critical edge systems. In this work, we propose a latency-oriented neural network learning method to optimize models for high accuracy while fulfilling the latency constraint. For efficiency, we also introduce a universal hardware-customized latency predictor to optimize this procedure to learn a model that satisfies the latency constraint by only a one-shot training process. The experiment results reveal that, compared to state-of-the-art methods, our approach can well-fit the ‘hard’ latency constraint and achieve high accuracy. Under the same training settings as the original model and satisfying a 34 ms latency constraint on the ImageNet-100 dataset, we reduce GoogLeNet’s latency from 40.32 ms to 34 ms with a 0.14% accuracy reduction on the NVIDIA Jetson Nano. When coupled with quantization, our method can be further improved to only 0.04% drop for GoogLeNet. On the NVIDIA Jetson TX2, we compress VGG-19 from 119.98 ms to 34 ms and even improve its accuracy by 0.5%, and we scale GoogLeNet up from 20.27 ms to 34 ms and achieve higher accuracy by 0.78%. We also open source this framework at https://github.com/ntuliuteam/ZeroBN." @default.
- W4313288463 created "2023-01-06" @default.
- W4313288463 creator A5003986323 @default.
- W4313288463 creator A5024084621 @default.
- W4313288463 creator A5038598144 @default.
- W4313288463 creator A5046896680 @default.
- W4313288463 creator A5067594005 @default.
- W4313288463 creator A5069119264 @default.
- W4313288463 creator A5086240263 @default.
- W4313288463 date "2023-05-01" @default.
- W4313288463 modified "2023-09-29" @default.
- W4313288463 title "Latency-constrained DNN architecture learning for edge systems using zerorized batch normalization" @default.
- W4313288463 cites W2113312949 @default.
- W4313288463 cites W2117539524 @default.
- W4313288463 cites W2736953746 @default.
- W4313288463 cites W2892341857 @default.
- W4313288463 cites W2893813411 @default.
- W4313288463 cites W2962697884 @default.
- W4313288463 cites W3034609471 @default.
- W4313288463 cites W3035442140 @default.
- W4313288463 cites W3165698711 @default.
- W4313288463 cites W4283815570 @default.
- W4313288463 doi "https://doi.org/10.1016/j.future.2022.12.021" @default.
- W4313288463 hasPublicationYear "2023" @default.
- W4313288463 type Work @default.
- W4313288463 citedByCount "1" @default.
- W4313288463 countsByYear W43132884632023 @default.
- W4313288463 crossrefType "journal-article" @default.
- W4313288463 hasAuthorship W4313288463A5003986323 @default.
- W4313288463 hasAuthorship W4313288463A5024084621 @default.
- W4313288463 hasAuthorship W4313288463A5038598144 @default.
- W4313288463 hasAuthorship W4313288463A5046896680 @default.
- W4313288463 hasAuthorship W4313288463A5067594005 @default.
- W4313288463 hasAuthorship W4313288463A5069119264 @default.
- W4313288463 hasAuthorship W4313288463A5086240263 @default.
- W4313288463 hasConcept C108583219 @default.
- W4313288463 hasConcept C111919701 @default.
- W4313288463 hasConcept C113775141 @default.
- W4313288463 hasConcept C138236772 @default.
- W4313288463 hasConcept C154945302 @default.
- W4313288463 hasConcept C2776214188 @default.
- W4313288463 hasConcept C41008148 @default.
- W4313288463 hasConcept C50644808 @default.
- W4313288463 hasConcept C76155785 @default.
- W4313288463 hasConcept C79403827 @default.
- W4313288463 hasConcept C79974875 @default.
- W4313288463 hasConcept C82876162 @default.
- W4313288463 hasConcept C97385483 @default.
- W4313288463 hasConceptScore W4313288463C108583219 @default.
- W4313288463 hasConceptScore W4313288463C111919701 @default.
- W4313288463 hasConceptScore W4313288463C113775141 @default.
- W4313288463 hasConceptScore W4313288463C138236772 @default.
- W4313288463 hasConceptScore W4313288463C154945302 @default.
- W4313288463 hasConceptScore W4313288463C2776214188 @default.
- W4313288463 hasConceptScore W4313288463C41008148 @default.
- W4313288463 hasConceptScore W4313288463C50644808 @default.
- W4313288463 hasConceptScore W4313288463C76155785 @default.
- W4313288463 hasConceptScore W4313288463C79403827 @default.
- W4313288463 hasConceptScore W4313288463C79974875 @default.
- W4313288463 hasConceptScore W4313288463C82876162 @default.
- W4313288463 hasConceptScore W4313288463C97385483 @default.
- W4313288463 hasLocation W43132884631 @default.
- W4313288463 hasOpenAccess W4313288463 @default.
- W4313288463 hasPrimaryLocation W43132884631 @default.
- W4313288463 hasRelatedWork W1530536511 @default.
- W4313288463 hasRelatedWork W2126887587 @default.
- W4313288463 hasRelatedWork W2741836081 @default.
- W4313288463 hasRelatedWork W2766146978 @default.
- W4313288463 hasRelatedWork W3082895349 @default.
- W4313288463 hasRelatedWork W3136021864 @default.
- W4313288463 hasRelatedWork W3165698711 @default.
- W4313288463 hasRelatedWork W3203962812 @default.
- W4313288463 hasRelatedWork W4283030313 @default.
- W4313288463 hasRelatedWork W4327774331 @default.
- W4313288463 hasVolume "142" @default.
- W4313288463 isParatext "false" @default.
- W4313288463 isRetracted "false" @default.
- W4313288463 workType "article" @default.