Every night, radio telescopes bring millions of images of distant galaxies. You can't handle such a flood manually. Edwin Hubble was already sorting galaxies by shape, but today smart algorithms recognize them faster. The problem is algorithms need many examples labeled by humans.
The STRADAViT method solves it differently — like teaching a child to recognize a cat by showing it from different angles. The computer is also shown the same galaxies, rotated and noised. Instead of memorizing brightness, it grasps the pattern — curves, jets, clouds of dust.
STRADAViT turned out more accurate than many analogues and requires fewer resources. It can be applied to different telescopes. This will help automatically track how billions of galaxies changed over time. And the most astonishing object found this way is the giant radio galaxy J1420-0545: its size is 16 million light-years. If it were visible to the eye, it would cover an area of five full moons in the sky. This approach continues the work accidentally started by Karl Jansky in 1931.
STRADAViT doesn't just compare pictures — it learns by imagining what the same galaxy looks like from different sides, as if picturing its three-dimensional shape.