Beginning of this page.
Displaying present location on the site.
  1. Home
  2. About NEC
  3. NEC Channel! MiTA TV
  4. NEC's RAPID machine learning supports human decisions with artificial intelligence, improving work efficiency!
Main content starts here.

NEC's RAPID machine learning supports human decisions with artificial intelligence, improving work efficiency!

Have you all heard the term "artificial intelligence"? It refers to the artificial computer-based realization of human intelligence, and this has recently become a topic discussed with great interest on TV and the Internet.

Apparently, NEC has developed software that incorporates one of these artificial intelligence technologies. Namely, the RAPID machine learning technology we'll discuss in this installment! This software enables you to have computers make human-like decisions by training them with data such as images and text. They say that having computers take over or support tasks where people have to make some kind of decision will lead to better work efficiency. Doesn't futuristic technology like this sound exciting? I wasted no time in going to interview the people in charge to hear more details!

Interviewee: Yasuyuki Tomonaga
Mr. Tomonaga is a software engineer engaged in the development and maintenance of RAPID machine learning. He is also responsible for the planning and sales promotion of human resource matching solutions that utilize RAPID machine learning.

Interviewee: Ms. Kyoko Fujishige
Ms. Fujishige is a software engineer engaged in the development and maintenance of RAPID machine learning. She is also responsible for the sales promotion of RAPID machine learning in a wide range of fields.

Machine learning software with artificial intelligence as its core engine

  • MitaHello, Ms. Fujishige and Mr. Tomonaga. I'm very interested to hear about RAPID machine learning, but I'm not very familiar with "machine learning" as a term. Can you walk me through it step by step?
  • TomonagaAllow me to give a brief explanation of machine learning, then. I actually have a perfect anecdote.
  • MitaWhat sort of anecdote?
  • TomonagaAn acquaintance of mine has a son who will be two-years-old soon, and the other day he was showing him an apple, tomato, and strawberry, and teaching him their names. Over the course of showing his son the apple and telling him what it was a few times, the boy started to say "apple," every time he was shown something red. In other words, he learned its name. Machine learning involves carrying out this process with computers.
  • MitaI see. So having computers learn in the same way that humans do is "machine learning," huh?
  • TomonagaThat's right. Then, when he later showed the boy a strawberry and told him its name, he could tell strawberries and apples apart from their difference in size. However, when showing the boy the apple and tomato, it was hard for him to tell the difference because they look so similar. Therefore, he told his son that tomatoes have a green stem, and advised him to look at that feature to know a tomato from an apple. That enabled the boy to tell the difference between tomatoes and apples.
  • MitaSo he directed his son's attention to a feature and told him to judge based on that.
  • TomonagaExactly. With conventional machine learning, specialists input detailed data into computers to perform feature learning like this. This was an extremely time-consuming task. However, technology that removes the need for this time-consuming process has emerged. This technology is called "deep learning."
  • MitaDeep learning?
  • TomonagaDeep learning is a technology that enables machines to recognize things. But instead of providing features like the example of the boy from earlier, decision-making is facilitated by feeding computers a large amount of sample data, identifying objects like tomatoes, apples, and strawberries, and having the computer itself extract the features automatically. For example, if you were to show a computer an image of a cat and teach it that it was a cat, the computer would independently find features such as the size and position of its eyes, the length of its whiskers, and the shape of its ears. Then, when it is shown another image, it will be able to determine whether or not it is a cat.
  • MitaAmazing! That's almost like a person!
  • TomonagaIt really is like a person, and we consider this deep learning to be a type of artificial intelligence technology. However, because deep learning involves feeding the computer a large amount of sample data, it took an enormous amount of time to carry out. RAPID machine learning is the result of NEC independently tuning this deep learning technology so it can be run faster, using less resources, and with more accuracy.
  • MitaInteresting. Now I see!

Larger viewNEC performed independent tuning of deep learning. This system determines data quickly, with minimal resources, and high accuracy.

Menu starts here.

NEC Channel! MiTA TV

End of menu.

Top of this page