7 Making the Edge work through AI

 

This chapter covers

  • How to effectively partition work on edge and cloud systems.
  • Situational awareness applications.
  • Machine Learning and AI applications on edge systems.
  • Cloud training and edge inference.
  • Time-series analysis.
  • Federated computing – blending many edge systems into AI and decision systems.

So far, we have explored the hardware and software frameworks for edge computing.  We have not detailed exactly what edge systems do, however.  We should ask ourselves, “how does an edge device communicate with the cloud (or should it)?”  We should think about what actions does an edge computer make.  Logic and applications running on edge systems can be simple utilities and drivers for hardware and sensors simply to rely data back to the cloud.  At the other extreme, we even have edge systems make decisions themselves using machine learning.  This chapter details how edge systems interact with cloud systems.  We will examine what happens in the cloud when edge components are added to the mix.  We also will talk about cloud usage in machine learning and predictive analytics.

7.1 The Purpose for Clouds with Edge Computing

7.2 Work on the Cloud and the Edge

7.3 Edge Workloads

7.3.1 Edge Patterns

7.4 Example Workload Organization

7.4.1 Situational Awareness Applications

7.4.2 Machine Learning for the Edge

7.4.3 Rules and Decision Systems

7.4.4 Time-Series Analysis

7.4.5 Proportional Integral Derivative Controllers

7.4.6 Probabilistic Analysis Systems

7.4.7 Deep Learning Models

7.5 Federated Machine Learning

7.6 Training in the Cloud, Inference at the Edge

7.7 Proper Use of Machine Learning

7.8 Summary