Matches in SemOpenAlex for { <https://semopenalex.org/work/W4292829073> ?p ?o ?g. }
Showing items 1 to 73 of
73
with 100 items per page.
- W4292829073 abstract "Neural Architecture Search (NAS) has enabled automatic discovery of more efficient neural network architectures, especially for mobile and embedded vision applications. Although recent research has proposed ways of quickly estimating latency on unseen hardware devices with just a few samples, little focus has been given to the challenges of estimating latency on runtimes using optimized graphs, such as TensorRT and specifically for edge devices. As devices like NVIDIA’s Jetsons get more popular in embedded computing and robotics, we observe a pressing need to more accurately estimate inference latency of neural network architectures on diverse runtimes, including highly optimized ones. In this work, we propose MAPLE-Edge, an edge device-oriented extension of MAPLE, the state-of-the-art latency predictor for general purpose hardware, where we train a regression network on architecture-latency pairs in conjunction with a hardware-runtime descriptor to effectively estimate latency on a diverse pool of edge devices. Compared to MAPLE, MAPLE-Edge can describe the runtime and target device platform using a much smaller set of CPU performance counters that are widely available on all Linux kernels, while still achieving up to +49.6% accuracy gains against previous state-of-the-art baseline methods on optimized edge device runtimes, using just 10 measurements from an unseen target device. We also demonstrate that unlike MAPLE which performs best when trained on a pool of devices sharing a common runtime, MAPLE-Edge can effectively generalize across runtimes by applying a trick of normalizing performance counters by the operator latency, in the measured hardware-runtime descriptor. Lastly, we show that for runtimes exhibiting lower than desired accuracy, performance can be boosted by collecting additional samples from the target device, with an extra 90 samples translating to gains of nearly +40%." @default.
- W4292829073 created "2022-08-24" @default.
- W4292829073 creator A5000469793 @default.
- W4292829073 creator A5034161060 @default.
- W4292829073 creator A5040284711 @default.
- W4292829073 creator A5082474713 @default.
- W4292829073 date "2022-06-01" @default.
- W4292829073 modified "2023-10-16" @default.
- W4292829073 title "MAPLE-Edge: A Runtime Latency Predictor for Edge Devices" @default.
- W4292829073 cites W2194775991 @default.
- W4292829073 cites W2963150697 @default.
- W4292829073 cites W2963918968 @default.
- W4292829073 cites W2967733054 @default.
- W4292829073 cites W2980137827 @default.
- W4292829073 cites W3034609471 @default.
- W4292829073 cites W3165698711 @default.
- W4292829073 cites W3214953945 @default.
- W4292829073 doi "https://doi.org/10.1109/cvprw56347.2022.00410" @default.
- W4292829073 hasPublicationYear "2022" @default.
- W4292829073 type Work @default.
- W4292829073 citedByCount "2" @default.
- W4292829073 countsByYear W42928290732023 @default.
- W4292829073 crossrefType "proceedings-article" @default.
- W4292829073 hasAuthorship W4292829073A5000469793 @default.
- W4292829073 hasAuthorship W4292829073A5034161060 @default.
- W4292829073 hasAuthorship W4292829073A5040284711 @default.
- W4292829073 hasAuthorship W4292829073A5082474713 @default.
- W4292829073 hasBestOaLocation W42928290732 @default.
- W4292829073 hasConcept C111919701 @default.
- W4292829073 hasConcept C118524514 @default.
- W4292829073 hasConcept C138236772 @default.
- W4292829073 hasConcept C154945302 @default.
- W4292829073 hasConcept C162307627 @default.
- W4292829073 hasConcept C173608175 @default.
- W4292829073 hasConcept C2780414537 @default.
- W4292829073 hasConcept C41008148 @default.
- W4292829073 hasConcept C50644808 @default.
- W4292829073 hasConcept C59822182 @default.
- W4292829073 hasConcept C76155785 @default.
- W4292829073 hasConcept C79974875 @default.
- W4292829073 hasConcept C82876162 @default.
- W4292829073 hasConcept C86803240 @default.
- W4292829073 hasConceptScore W4292829073C111919701 @default.
- W4292829073 hasConceptScore W4292829073C118524514 @default.
- W4292829073 hasConceptScore W4292829073C138236772 @default.
- W4292829073 hasConceptScore W4292829073C154945302 @default.
- W4292829073 hasConceptScore W4292829073C162307627 @default.
- W4292829073 hasConceptScore W4292829073C173608175 @default.
- W4292829073 hasConceptScore W4292829073C2780414537 @default.
- W4292829073 hasConceptScore W4292829073C41008148 @default.
- W4292829073 hasConceptScore W4292829073C50644808 @default.
- W4292829073 hasConceptScore W4292829073C59822182 @default.
- W4292829073 hasConceptScore W4292829073C76155785 @default.
- W4292829073 hasConceptScore W4292829073C79974875 @default.
- W4292829073 hasConceptScore W4292829073C82876162 @default.
- W4292829073 hasConceptScore W4292829073C86803240 @default.
- W4292829073 hasLocation W42928290731 @default.
- W4292829073 hasLocation W42928290732 @default.
- W4292829073 hasOpenAccess W4292829073 @default.
- W4292829073 hasPrimaryLocation W42928290731 @default.
- W4292829073 hasRelatedWork W2117014006 @default.
- W4292829073 hasRelatedWork W2372170743 @default.
- W4292829073 hasRelatedWork W2736305332 @default.
- W4292829073 hasRelatedWork W3140030527 @default.
- W4292829073 hasRelatedWork W3184768109 @default.
- W4292829073 hasRelatedWork W4213074318 @default.
- W4292829073 hasRelatedWork W4233815414 @default.
- W4292829073 hasRelatedWork W4281678247 @default.
- W4292829073 hasRelatedWork W4287076991 @default.
- W4292829073 hasRelatedWork W4376106090 @default.
- W4292829073 isParatext "false" @default.
- W4292829073 isRetracted "false" @default.
- W4292829073 workType "article" @default.