The USA and Russia develop neural networks intended to prevent natural disasters, Yandex improves Soviet cartoons using AI, and a picture written by AI is exhibited in New York. Read about these and other AI news in our weekly digest.
Google’s AI predicts typhoons
Google developed a neural network capable of predicting natural disasters. Google plans to place forecasts near the search line and in Google Maps.
Developers taught the neural network to forecast floods and typhoons based on the set of data from historical events to the water level in rivers. AI also factors in weather forecasts and statistical data.
The new project will be tested in Patna, India, as part of Google’s partnership with the country’s Central Water Commission. The first warnings about disasters are expected to appear in October.
Neural network vs earthquakes
Russian scientists of the National Research Tomsk State University (TSU) together with colleagues from the University of Tokyo are developing a neural network that can forecast earthquakes, landslides, and mudflows.
According to TSU, the project will help to prevent negative consequences of landslides, coastal destruction, and flooding of cities and villages. Rescue services are expected to use the forecasts in order to conduct planned activities and take preventive measures.
Medical database developed in Russia
Association of artificial intelligence developers and participants of insurance and medical communities plan to develop a single base containing biomedical data.
This database is supposed to become a basis for teaching different neural networks to make correct diagnoses. The head of digital medicine department at Skolkovo Sergey Voinov stated that the creation of the database on the most common diseases would take up to five years. In theory, such kind of data can be useful for insurance companies that can deploy data to assess customer risks.
The government got interested in the project, as it will allow building a reliable medical database for the health care system.
A new payment system for public transport developed in Russia
Ministry of Construction Industry, Housing and Utilities Sector of Russia (Minstroy) plans to integrate the face recognition system in public transport to charge fees directly from bank accounts of passengers.
The deputy minister of Minstroy Andrey Chibis believes that boarding lingers when passengers use transport cards or cash. Face recognition system will help citizens to reach the destination quicker and spend less time on the purchase of tickets.
The system will be developed by two companies – Russian telephony provider Rostelecom and Chinese Huawei.
Painting drawn by AI exhibited at the auction
On October 23-25, Christie’s auction house will be selling the Portrait of Edmond Belamy drawn by the artificial intelligence algorithm.
AI algorithm was developed by the French developers of The Obvious team. The person depicted in the picture is an imaginary member of the French family Belamy. One of the expert auctioneers states that although the portrait was created by artificial intelligence, it is a real piece of art like all paintings sold by Christie’s within two centuries and a half.
The Obvious wants to receive around €10,000 for the painting. Funds will be used to improve the artificial intelligence algorithm.
Yandex AI upgraded old cartoons
Yandex used artificial intelligence to improve the quality of image and resolution of tens of Soviet cartoons including the Bremen Town Musicians, the Gold Antelope, the Thumbelina, the Snow Queen, and others.
To process the video, Yandex used its own development based on neural networks, the DeepHD technology. It can increase the resolution, remove drawbacks and other graphic artefacts, and improve detalization and sharpness.
The upgraded cartoons are already available on the Internet.