Published nearly six months ago, this is a good article...

Published nearly six months ago, this is a good article detailing the position of AI at the Edge and the synergistic relationship with Cloud-based AI.

Considering the referenced example for algorithm development the article also highlights just how far ahead Akida is with regards to continuous learning and object recognition.

One-shot learning demonstrated here:



https://www.forbes.com/sites/mohanbirsawhney/2020/01/27/why-apple-and-microsoft-are-moving-ai-to-the-edge/#7103310a2570

8,306 views|Jan 27, 2020,12:33pm EST

Why Apple And Microsoft Are Moving AI To The Edge

Mohanbir SawhneyContributor

CMO Network

I analyze trends and current eventsin technology, marketing and AI.

Artificial intelligence (AI) has traditionally been deployed inthe cloud, because AI algorithms crunch massive amounts of data and consumemassive computing resources. But AI doesn’t only live in the cloud. In manysituations, AI-based data crunching and decisions need to be made locally, ondevices that are close to the edge of the network.

AI at the edge allows mission-critical and time-sensitivedecisions to be made faster, more reliably and with greater security. The rushto push AI to the edge is being fueled by the rapid growth of smart devices atthe edge of the network—smartphones, smart watches and sensors placed onmachines and infrastructure. Earlier this month, Apple spent $200 million to acquire Xnor.ai, a Seattle-based AI startup focused on low-power machine learning software and hardware. Microsoft offers a comprehensive toolkit called Azure IoT Edge that allows AI workloads to be moved to the edge of the network.

Will AI continue to move to theedge? What are the benefits and drawbacks of AI at the edge versus AI in thecloud? To understand what the future holds for AI at the edge, it is useful tolook back at the history of computing and how the pendulum has swung fromcentralized intelligence to decentralized intelligence across four paradigms of computing.

Centralized vs. Decentralized

Since the earliest days of computing, one of the designchallenges has always been where intelligence should live in a network. As Iobserved in an article in the Harvard Business Review in 2001, there has been an “intelligence migration” from centralized intelligence to decentralized intelligence—a cycle that’s now repeating.

The first era of computing was themainframe, with intelligence concentrated in a massive central computer thathad all the computational power. At the other end of the network were terminalsthat consisted essentially of a green screen and a keyboard with little intelligenceof their own—hence they were called “dumb terminals.”

The second era of computing was the desktop or personal computer(PC), which turned the mainframe paradigm upside down. PCs contained all theintelligence for storage and computation locally and did not even need to beconnected to a network. This decentralized intelligence ushered in the democratizationof computing and led to the rise of Microsoft and Intel, with the vision ofputting a PC in every home and on every desk.

The third era of computing, calledclient-server computing, offered a compromise between the two extremes ofintelligence. Large servers performed the heavy lifting at the back-end, and “front-end intelligence” was gathered and stored on networked client hardware and software.

The fourth era of computing is thecloud computing paradigm, pioneered by companies like Amazon with its AmazonWeb Services, Salesforce.com with its SaaS (Software as a Service) offerings,and Microsoft with its Azure cloud platform. The cloud provides massivelyscaled computational power and very cheap memory and storage. It only makessense that AI applications would be housed in the cloud, since the computationpower of AI algorithms has increased 300,000 times between 2012 and 2019—doubling every three-and-a-half months.

The PendulumSwings Again

Cloud-based AI, however, has itsissues. For one, cloud-based AI suffers from latency—the delay as data moves tothe cloud for processing and the results are transmitted back over the networkto a local device. In many situations, latency can have serious consequences.For instance, when a sensor in a chemical plant predicts an imminent explosion,the plant needs to be shut down immediately. A security camera at an airport ora factory must recognize intruders and react immediately. An autonomous vehiclecannot wait even for a tenth of a second to activate emergency braking when theAI algorithm predicts an imminent collision. In these situations, AI must belocated at the edge, where decisions can be made faster without relying onnetwork connectivity and without moving massive amounts of data back and forthover a network.

The pendulum swings again, fromcentralization to decentralization of intelligence— just as we saw 40 years agowith the shift from mainframe computing to desktop computing.

However, as we found out with PCs,life is not easy at the edge. There is a limit to the amount of computationpower that can be put into a camera, sensor, or a smartphone. In addition,many of the devices at the edge of the network are not connected to a power source,which raises issues of battery life and heat dissipation. These challenges arebeing dealt with by companies such as Tesla, ARM, and Intel as they developmore efficient processors and leaner algorithms that don’t use as much power.

But there are still times when AI isbetter off in the cloud. When decisions require massive computational power and do not need to be made in real time, AI should stay in the cloud. For example, when AI is used to interpret an MRI scan or analyze geospatial data collected by a drone over a farm, we can harness the full power of the cloud even if we have to wait a few minutes or a few hours for the decision.

Training vs.Inference

One way to determine where AI shouldlive is to understand the difference between training and inference in AIalgorithms. When AI algorithms are built and trained, the process requiresmassive amounts of data and computational power. To teach an autonomous vehicleto recognize pedestrians or stop lights, you need to feed the algorithmmillions of images. However, once the algorithm is trained, it can perform“inference” locally—looking at one object to determine if it is a pedestrian.In inference mode, the algorithm leverages its training to make lesscomputation-intensive decisions at the edge of the network.

AI in the cloud can worksynergistically with AI at the edge. Consider an AI-powered vehicle like Tesla.AI at the edge powers countless decisions in real time such as braking,steering, and lane changes. At night, when the car is parked and connected to aWi-Fi network, data is uploaded to the cloud to further train the algorithm.The smarter algorithm can then be downloaded to the vehicle over the cloud—avirtuous cycle that Tesla has repeated hundreds of time through cloud-basedsoftware updates.

Embracingthe Wisdom of the “And”

There will be a needfor AI in the cloud, just as there will be more reasons to put AI at the edge.It isn’t an either/or answer, it’s an “and.” AI will be where it needs to be,just as intelligence will live where it needs to live. I see AI evolving into“ambient intelligence”—distributed, ubiquitous, and connected. In this visionof the future, intelligence at the edge will complement intelligence in thecloud, for better balance between the