The "Neural Debris Removal in Streak Detection Models" competition, held under the patronage of the European Space Agency (ESA), ran from April 22 to July 22, 2026 and involved 560 young researchers (students and PhD candidates) from all over the world.
The challenge its participants faced was to identify and neutralize a form of "poison" deliberately introduced into an AI model - which caused the Artificial Intelligence to misidentify objects in images captured by telescopes.
Over the course of three months, participants were required to submit successive proposals to solve the problem. The one presented by the Polish engineer, supported by her fellow student Michał Duszak, succeeded in achieving a high fifth place overall.
The young Polish scientists' solution developed for the competition was presented in September at the KGK Space Resources Conference held at the AGH University of Science and Technology in southern Poland's city of Kraków.
As noted by the Opole University of Technology press office, Magdalena Machacka's achievement marks yet another success for the talented researcher, following her highly acclaimed engineering thesis on a Polish Sign Language recognition system based on an Artificial Intelligence model.
(mm)
Source: PAP