In a new paper with Max, Lukas and Günther, recently published in Nature Photonics, we explored how artificial intelligence can extract the maximum possible information from distorted optical signals. Using neural networks trained on seemingly chaotic light patterns, we approached — and nearly reached — the theoretical limit of precision, which we showed to be defined by the Cramér–Rao bound. Many thanks go to our collaborators in France (Dorian Bouchet) and Scotland (Ilya Starshynov and Daniele Faccio) as well as to Oliver for creating the great feature image shown on the top left. Check here for TU Wien’s press release.
