JAPANESE
CONTACT

Machine learning that expands the base

Electronic payments have spread throughout the city, and QR code tags are now lined up at store cashiers. It is an interface for electronic money. This is a method in which the user reads the QR code with his or her smartphone, enters the payment amount, and makes an electronic payment. There is also a method of making an electronic payment by having the cashier machine read the QR code or barcode displayed on the smartphone, but it is a little safe because you enter the payment amount yourself.

By the way, until a while ago, I thought that it was easy to use electronic payments using cards with built-in IC chips and mobile phones, but for not a few stores now it is only a method of reading the QR code placed in the store. I think that the number of stores has increased and the flow of electronic payment has changed somewhat. Among them, the reading accuracy by the smartphone camera has improved dramatically. It seems that the results of machine learning are being used skillfully. A few years ago, the QR code was read by the mobile camera according to the guide displayed on the mobile screen, but now if the QR code is in the field of view of the smartphone camera, smartphone application will find the QR code regardless of the location, size and tilt of the code. Once found, it will decrypt the code. The smartphone app does not do machine learning but uses the results of machine learning.

Modeling machine learning requires a large number/amount of samples and computer power, but once the model and parameters are determined, it seems that recognition is performed according to the model with relatively small computer power.

Many apps that use the camera installed in the smartphone use this recognition mechanism. There are a wide variety of apps, such as apps that tell you the names of flowers, and apps that search the internet for similar things. It has become very convenient, but on the other hand, I think it has become exceedingly difficult to ensure privacy.

 

Nobuo FUJII was working for NTT R&D and NTT Groups where he studied and developed transport network operation systems. He also worked for ITU-T and other standard organizations, fora and consortia so as to develop operation system architectures and interfaces. He is an IEICE fellow. He joined CCI in 2014.

Other Posts