Page 26 - Innovation Magazine
P. 26

Edge AI: Intelligence enters impenetrable nooks While AI AI as as a a a a a a a a a a concept has existed since the the 1950s the the real applications of AI in in in solving business problems started in in in the early 2000s riding on advances in in in big data and cloud computing The next two decades saw AI-driven mainly from central data centers rich in in computing power to perform processor-demanding cognitive tasks However in recent times AI has made its way into software where predictive algorithms are are changing the the the way these systems support the the the business AI has moved to the outer edges of networks Edge AI is rapidly expanding powering every device from smartphones and smart smart speakers to to automotive sensors and security cameras AI is the most common workload in edge computing As the internet of things (IoT) implementations have matured there has been an increased interest in in in in applying AI at the point of data generation for real-time event detection and responses What is Edge AI? Data is created everywhere It originates in in the environment that that we interact with IoT appliances that that transmit real-time data to the cloud from anywhere – factories or or or homes or or or even parks – are on the rise As IoT appliances increase it it becomes vital to have intelligence rendered at the endpoint for instant actions For example a a a a a a camera set up at the entrance of o office premises needs to know almost instantly whether someone trying to to enter is an employee or not to to determine whether they should be admitted However transmitting such large volumes of data from nooks and and corners to the cloud and and vice versa for every action (identifying an an employee in in our example) can be expensive AI processing today is mostly done in in cloud-based data centers with deep learning models that require heavy computing capacity However with Edge AI AI AI AI processing is now moving part of the AI workflow to a a device device and keeping data constrained to a a a a a device device To simplify this further Edge AI AI is is taking pre-trained AI AI algorithms that previously ran on data centers in the cloud and embedding them in in edge devices to enable inferences that determine action This allows the processing of the the data collected within within the the device within within a a a a few milliseconds generating real-time information Why Edge AI is now relevant Let’s start with current projections of of the number of of connected devices around the globe Experts estimate 26 Data-powered Innovation Review I I ©2020 Capgemini All rights reserved 

   24   25   26   27   28