Scientists at the U.S. Department of Energy's Argonne National Laboratory have developed an AI system that can analyze X-ray images of materials and identify their structure, deformations, and defects. Thanks to this high-tech solution, work that previously took weeks can now be completed hundreds of times faster.
DONUT learns on the fly during experiments and, most importantly, does not require large labeled datasets. Instead, the neural network relies on the physical laws governing the interaction of X-rays with the specific material being studied.
In the coming years, this technology could become the backbone of autonomous laboratories, where artificial intelligence will independently adjust experimental conditions in real time.