Self-learning AI created in Kazakhstan

Scientists in Kazakhstan have created an artificial neural network capable of self-learning, a Tengrinews correspondent reports citing the director of the developer company Adil Augaliyev.

The system called iSam consists of 250 million neural connections. The artificial neural network is arranged following the principle of biological networks of nerve cells of living organisms. This system is capable of recognizing objects in high-definition images, learning the patters of the objects and then recognizing them elsewhere.

The project is similar to those attempted by Google and Microsoft.

The software developed in Kazakhstan can serve as a platform for a universal technology  that can be integrated into transport security systems, accounting and storage management. In addition, it can be used to study marketing potential. For example, in large stores, the neural network is capable of tracking the eye lines of customers and identifying the areas that attract the most attention. It can also identify the age and gender of the customers.

Creators say the program is 80 percent ready. They are already testing it. In the final version, it can be used in the Safe City project to integrate all the video surveillance systems into a single network. The artificial intelligence will be able to analyze not only routine events like traffic jams, but will also be capable of identifying potentially dangerous situations - such as riotous assemblies, unusual behavior, possession of alcoholic beverages, etc.

The developers have already presented their technology to the traffic police of Almaty and the latter expressed interest in the project.

By Dinara Urazova, editing by Tatyana Kuzmina