Mid Sweden University

miun.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Sound event detection with binary neural networks on tightly power-constrained IoT devices
Show others and affiliations
2020 (English)In: ACM International Conference Proceeding Series, Association for Computing Machinery , 2020, article id 3406588Conference paper, Published paper (Refereed)
Abstract [en]

Sound event detection (SED) is a hot topic in consumer and smart city applications. Existing approaches based on deep neural networks (DNNs) are very effective, but highly demanding in terms of memory, power, and throughput when targeting ultra-low power always-on devices. Latency, availability, cost, and privacy requirements are pushing recent IoT systems to process the data on the node, close to the sensor, with a very limited energy supply, and tight constraints on the memory size and processing capabilities precluding to run state-of-The-Art DNNs. In this paper, we explore the combination of extreme quantization to a small-footprint binary neural network (BNN) with the highly energy-efficient, RISC-V-based (8+1)-core GAP8 microcontroller. Starting from an existing CNN for SED whose footprint (815 kB) exceeds the 512 kB of memory available on our platform, we retrain the network using binary filters and activations to match these memory constraints. (Fully) binary neural networks come with a natural drop in accuracy of 12-18% on the challenging ImageNet object recognition challenge compared to their equivalent full-precision baselines. This BNN reaches a 77.9% accuracy, just 7% lower than the full-precision version, with 58 kB (7.2× less) for the weights and 262 kB (2.4× less) memory in total. With our BNN implementation, we reach a peak throughput of 4.6 GMAC/s and 1.5 GMAC/s over the full network, including preprocessing with Mel bins, which corresponds to an efficiency of 67.1 GMAC/s/W and 31.3 GMAC/s/W, respectively. Compared to the performance of an ARM Cortex-M4 implementation, our system has a 10.3× faster execution time and a 51.1× higher energy-efficiency. © 2020 ACM.

Place, publisher, year, edition, pages
Association for Computing Machinery , 2020. article id 3406588
Keywords [en]
binary neural networks, sound event detection, ultra low power, Deep neural networks, Energy efficiency, Internet of things, Low power electronics, Object recognition, Privacy by design, Energy efficient, Memory constraints, Privacy requirements, Processing capability, Small footprints, State of the art, Neural networks
Identifiers
URN: urn:nbn:se:miun:diva-41573DOI: 10.1145/3370748.3406588Scopus ID: 2-s2.0-85098283100ISBN: 9781450370530 (print)OAI: oai:DiVA.org:miun-41573DiVA, id: diva2:1536175
Conference
2020 ACM/IEEE International Symposium on Low Power Electronics and Design, ISLPED 2020; Virtual, Online; United States; 10 August 2020 through 12 August 2020
Available from: 2021-03-10 Created: 2021-03-10 Last updated: 2021-04-30Bibliographically approved

Open Access in DiVA

No full text in DiVA

Other links

Publisher's full textScopus

Search in DiVA

By author/editor
Magno, M.

Search outside of DiVA

GoogleGoogle Scholar

doi
isbn
urn-nbn

Altmetric score

doi
isbn
urn-nbn
Total: 5 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf