ABSTRACT
The past decade has seen machine learning transform from theory and ideation to real-life applications. Consumer industries like retail and e-commerce stores have started to learn consumer behaviour in real-time, the banking sector is relying upon machine learning models to spot hackers and frauds, and social media sites are supplying user-specific ads and content to keep users engaged on their platforms. While all these applications are very distinctive, they have two commonalities: enormous data generated every second and optimization in real-time to make the services better. The sheer dependence of these models to make quick and accurate decisions requires them to inherit the fundamentals of fog computing. Since it’s nearly impossible to collate and process data generated by these applications on the cloud, processing them on the edge using legacy data models becomes necessary for node-based result optimization. Industries like oil and gas, where time is of the essence and processing a single data point’s time could be a matter of either success or disaster, find fog computing at their core.
