Hlavní stránka Info portálu ZČU

Eduard Plic, a student at Masaryk High School in Pilsen, described in his paper the use of machine learning in sunspot tracking and classification. For this work he won first place in the SOČ in Physics and nomination to the international round. Under the supervision of Martin Bulín from the Department of Cybernetics, Faculty of Applied Sciences, he used a convolutional neural network for sunspot classification and worked with a sun drawing protocol. After dividing the spots into groups according to the international classification, he performed computer processing of the drawings and trained machine learning models that classified the spots themselves.

''Some may think that since we have much more modern and better quality images from satellites, there is no need to draw the Sun by hand. But that is a mistake. We need to have both the drawing and the satellite data to get an idea of what sunspots looked like when satellites didn't exist or didn't work accurately,'' says Eduard Plic. Continuously recording data using the same method, he says, allows us to understand what is going on inside the Sun and to predict threats such as charged particle bursts that can short out power lines. ''We could observe this phenomenon in the last century in the US,'' Eduard adds. 

In the Czech Republic, only a few observatories are dedicated to drawing the Sun, including the main observatory of the Astronomical Institute of the CAS in Ondřejov, where sunspots have been recorded daily since 1943. ''Machine learning cannot replace humans, but it can draw attention to unusual activity,'' emphasizes educator Martin Bulín, while praising the cooperation with the gifted high school student: ''The cooperation with Eduard was excellent and I believe he has a promising future in research.''