How can neural networks learn the structure of images, even when most of the information is missing? This is the central question of this PhD project. In tomographic imaging, the goal is to reconstruct a three-dimensional object from observations acquired along a limited number of directions. Such problems are inherently ill-posed, yet recent advances in machine learning have shown that it is possible to recover high-quality reconstructions even under severe degradation.
Please see the detailed subject in the attached PDF or at https://valentindebarnot.github.io/pages/OpenPositions.
