The World's Top AI Technology Is In Japan!
3 main points
✔️ Unique technologies of Japanese companies that are not found anywhere else in the world!
✔️ Enabling existing products to become highly intelligent
✔️ Be the leading edge of edge AI in the future
One-point explanation from AI-SCHOLAR
I think that in 2020 we were only exploring the scope of what AI could adapt to. In reality, there was no environment for AI, and implementation was difficult in some areas. However, making use of the experience of 2020, we are already making steady progress in the development of AI implementation for the future. One of them is edge AI. However, the memory problem, which is a major challenge, has been solved this time.
(AISing Ltd.), an edge AI startup announced on December 15 that it has developed a Memory Saving Tree (MST), an edge AI algorithm that is so small it can be placed on a fingertip.
For those of you who don't understand, I'd like to summarize the background, the main points, and what this technology can do for you.
The key is edge AI and microcontrollers
I'll explain everything, don't worry!
First of all, there are two main types of AI: AI processing in the cloud (cloud AI) and AI processing on the edge (edge AI). To put it simply, cloud AI can be used as long as communication is available, and edge AI can be used regardless of communication.
It would be nice if there were no problems whatsoever with the cloud, but there aren't. Cloud AI is useless on a fishing boat or deep in the mountains because you need to have a good communication environment. In addition, there are many things in the world that can be fatal if communication is interrupted or delayed.
For example, imagine this. If automated driving were to slow down to recognize the human in front of it due to communication delays when it's being processed by cloud AI, you'd be horrified.
With that being said, edge AI is in much demand.
But here's the big problem: AI requires a lot of computing power, so to make AI usable for existing, you need big hardware. Basically, we want machines to be small with high performance. But the reality is difficult, and there are general tradeoffs, such as high performance means bigger, and low performance means smaller, so you can't implement AI in everything. That's why edge AI has been slow to develop.
Think of it as a microcontroller, a tiny brain in an electronic device. It is always contained in home appliances such as refrigerators, washing machines, and rice cookers. In the case of a rice cooker, the timer and display functions are also performed by a tiny microcontroller. As far as rice cookers are concerned, you have succeeded in miniaturization with high performance.
In addition, it is important to remember that Arm-based microcontrollers are installed in the majority of the world's devices, and Apple and Amazon are among the GAFAs that are paying attention to Arm, and it is said to grow the most in 2021, so it is important to remember the word Arm.
It's becoming commonplace for existing devices to become highly intelligent, and edge AI is getting a lot of attention, but AI is too big to be considered hard to implement in existing products.
Memory Saving Tree (MST)
The Memory Saving Tree (MST), as it turns out, has solved this problem!
Small enough to be placed on your fingertips Ultra small Ultra small AI that can be implemented on it. → High intelligence High intelligence (same as offline) Online learning Further online learning → Accuracy Accuracy is getting better and better In addition, it covers about 92% of Arm-based microcontrollers. In addition, about 92% of Arm-based microcontrollers are covered. → Get on the existing products .
I think you can see that this will have too much of an impact. It will be the kind of technology that even that GAFA hasn't done, and maybe can't do.
Home appliances may be affected tremendously. All home appliances will be equipped with AI, and while processing the information on the edge, it may even go so far as to take into account a person's preferences and health.
This means that there is now a possibility that AI can be implemented in areas that were previously thought to be impossible. New businesses may be created by introducing AI into areas that have never been adapted before.
And the fact that it can be used in existing products means that manufacturers of electronic devices will be able to make their products more intelligent. In addition, there will be further development by making use of the information from edge AI, which will differentiate us from our competitors.
From our point of view, not having AI onboard may be as good as simply not having one feature.
We hope you'll catch up with this article and remember these words.
AISing, MST, Microcontroller (Arm)