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/

Global: Zendy Plus annual

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

Global: Zendy Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Human Activity Recognition (HAR) is a relevant inference task in many mobileapplications. State-of-the-art HAR at the edge is typically achieved withlightweight machine learning models such as decision trees and Random Forests(RFs), whereas deep learning is less common due to its high computationalcomplexity. In this work, we propose a novel implementation of HAR based ondeep neural networks, and precisely on Binary Neural Networks (BNNs), targetinglow-power general purpose processors with a RISC-V instruction set. BNNs yieldvery small memory footprints and low inference complexity, thanks to thereplacement of arithmetic operations with bit-wise ones. However, existing BNNimplementations on general purpose processors impose constraints tailored tocomplex computer vision tasks, which result in over-parametrized models forsimpler problems like HAR. Therefore, we also introduce a new BNN inferencelibrary, which targets ultra-compact models explicitly. With experiments on asingle-core RISC-V processor, we show that BNNs trained on two HAR datasetsobtain higher classification accuracy compared to a state-of-the-art baselinebased on RFs. Furthermore, our BNN reaches the same accuracy of a RF witheither less memory (up to 91%) or more energy-efficiency (up to 70%), dependingon the complexity of the features extracted by the RF.

Ultra-compact Binary Neural Networks for Human Activity Recognition on RISC-V Processors