All news

Russian students teach ANN to forecast ice traffic in Arctic

The system will capture patterns of interaction between close ice areas and the influence of weather, thus predicting the ice dynamics

BARNAUL, September 7. /TASS/. Russian students, winners in an artificial intelligence (AI) competition, taught an artificial neural network (ANN) to assess ice traffic in the Arctic in the interests of the Northern navigation, Skoltech's press service told TASS.

"Anatoly Onishchenko from Yekaterinburg and Stefan Maria Ailuro from Moscow have offered a model to forecast ice traffic in a bay," the press service said. "This model can improve navigation safety in the North where satellite data are not sufficient."

According to the developers, the project will be helpful for sailors who need to understand how ice is moving. "Normally, they would refer to satellite images, but such images are not daily, and thus classical machine learning methods are not applicable here," the press service quoted the students as saying.

The system will capture patterns of interaction between close ice areas and the influence of weather, thus predicting the ice dynamics. The model may be used if real data are not sufficient. The developers have been offered an internship at the Skoltech Applied AI Center, where they will be able to continue working on a project to combine computational models and AI in Arctic ice forecasts.

The project is a winner in a hackathon during the summer school, which was held for 12 days under the Artificial Intelligence federal project of the Digital Economy National Project. The event featured 65 students, masters and postgraduates from Russian universities involved in AI research, and more than 200 people joined the school online.