research1d
Embedded Machine Learning for Microcontroller-Class Edge Devices: Data, Feature, Evaluation, and Deployment Pipelines
This paper presents a systems-oriented approach to embedded machine learning on microcontroller-class edge devices, focusing on the engineering decisions involved in deploying machine learning models on resource-constrained platforms. The workflow covers data acquisition, preprocessing, feature extraction, model deployment, and inference. The authors provide a detailed evaluation of the design choices and their impact on memory, energy, and latency. You can use this approach to optimize your own
Key takeaways
- Emphasizes engineering decisions for embedded machine learning on microcontrollers
- Covers data acquisition, preprocessing, feature extraction, and model deployment
- Provides a detailed evaluation of design choices and their impact on memory, energy, and latency