Bringing Intelligence to Edge Computing Through Machine Learning
Machine learning is one of the top use cases of edge computing. By detecting anomalies and patterns in data streams and initiating appropriate actions, machine learning models support augmented and virtual reality, connected vehicles, industrial IoT, smart cities, smart grids, and smart healthcare use cases. Low-powered IoT devices are alone not powerful enough to talk to public cloud platforms. Devices need to be connected through a gateway that acts as an edge computing layer. Backed by this layer, and graphics processing units, machine learning models are trained at scale in the cloud and pushed to the edge for inferencing (deducing).
This paper by Janakiram MSV, well-respected analyst, advisor, and cloud computing architect, explains how organizations can bring machine learning to the edge to enable intelligent IoT scenarios. It covers:
- Machine learning as the driver of edge computing
- The role of serverless computing
- The importance of interoperability among deep learning frameworks and the ONNX neural network
- Functions as a service and integration platforms as a service