English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

Optimizing resource utilization in target platforms is key to achieving highperformance during DNN inference. While optimizations have been proposed forinference latency, memory footprint, and energy consumption, priorhardware-aware neural architecture search (NAS) methods have omitted resourceutilization, preventing DNNs to take full advantage of the target inferenceplatforms. Modeling resource utilization efficiently and accurately ischallenging, especially for widely-used array-based inference accelerators suchas Google TPU. In this work, we propose a novel hardware-aware NAS frameworkthat does not only optimize for task accuracy and inference latency, but alsofor resource utilization. We also propose and validate a new computationalmodel for resource utilization in inference accelerators. By using the proposedNAS framework and the proposed resource utilization model, we achieve 2.8 - 4xspeedup for DNN inference compared to prior hardware-aware NAS methods whileattaining similar or improved accuracy in image classification on CIFAR-10 andImagenet-100 datasets.

U-Boost NAS: Utilization-Boosted Differentiable Neural Architecture Search