Machine Learning for Microcontroller-Class Hardware - A Review

The advancements in machine learning opened a new opportunity to bring intelligence to the low-end Internet-of-Things nodes such as microcontrollers. Conventional machine learning deployment has high memory and compute footprint hindering their direct deployment on ultra resource-constrained microco...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE sensors journal Jg. 22; H. 22; S. 1
Hauptverfasser: Saha, Swapnil Sayan, Sandha, Sandeep Singh, Srivastava, Mani
Format: Journal Article
Sprache:Englisch
Veröffentlicht: United States IEEE 15.11.2022
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Schlagworte:
Computational modeling
Data models
Feature projection
Hardware
Internet of Things
Machine learning
Mathematical models
Microcontrollers
model compression
neural architecture search
neural networks
optimization
Random access memory
Sensors
TinyML
Workflow
model compression
sensors
optimization
neural networks
Feature projection
internet-of-things
machine learning
TinyML
neural architecture search
microcontrollers
ISSN:1530-437X, 1558-1748
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The advancements in machine learning opened a new opportunity to bring intelligence to the low-end Internet-of-Things nodes such as microcontrollers. Conventional machine learning deployment has high memory and compute footprint hindering their direct deployment on ultra resource-constrained microcontrollers. This paper highlights the unique requirements of enabling onboard machine learning for microcontroller class devices. Researchers use a specialized model development workflow for resource-limited applications to ensure the compute and latency budget is within the device limits while still maintaining the desired performance. We characterize a closed-loop widely applicable workflow of machine learning model development for microcontroller class devices and show that several classes of applications adopt a specific instance of it. We present both qualitative and numerical insights into different stages of model development by showcasing several use cases. Finally, we identify the open research challenges and unsolved questions demanding careful considerations moving forward.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
content type line 23
ISSN:1530-437X
1558-1748
DOI:10.1109/JSEN.2022.3210773