Russian scientists, together with Austrian and American colleagues, have created a unique artificial intelligence system for processing long-term data about the Sun. The project was implemented by specialists from Skoltech, the University of Graz and the High Altitude Observatory of the USA.
Over the decades of observations, huge amounts of data have been accumulated from different telescopes. The problem is that due to differences in the quality of images and technical characteristics of the equipment, these data are difficult to compare. The new AI system solves this problem by combining information into a single model.
The technology is based on special neural networks (GAN) that work as a team of two specialists. One part of the system deliberately degrades the quality of images, and the second restores them to an ideal state. This approach allows you to correct images taken at different times and with different tools, as well as find hidden patterns.
AI has already processed data for 24 years of observations. The system has improved the clarity of images of the solar disk, reduced interference in images from Earth, and even learned to assess magnetic fields on the far side of the Sun. Scientists hope that the technology will help reveal new secrets about the connection between solar cycles and space weather.
Read more materials on the topic:
The sun arranged a fireworks display of 39 flares, almost half of them class M
Mars or the Moon?: a historian of astronautics told when Russia will be able to conquer Mars
It became known what will be installed on the Russian spacecraft "Luna-26" and "Luna-27"