Doctoral Researcher Siiri Rautio will defend her doctoral dissertation Ill-posed Microlocal Imaging Using Targeted Deep Learning in the Faculty of Science, University of Helsinki, on 3 April 2025 at 12:00.
As computational power has grown, so has the potential and significance of deep learning in all computational sciences.
In general, the more complicated the task is, the more data and learnable parameters the neural network needs to learn to solve it. In so-called targeted deep learning the task is decomposed into learned and non-learned parts, that are specifically tailored for the problem in hand. By only targeting a specific part of the problem with deep learning, the networks can be smaller and less expensive to train, and the end-result easier to interpret.
The field of inverse problems provides an interesting testing ground of problems for targeted deep learning to solve. With targeted deep learning, the ill-posed part of the problem can be solved with robust mathematics, providing the learned part with already more robust features.
Professor Elena Loli Piccolomini, University of Bologna, will serve as the opponent. FAME’s Deputy Vice Director Samuli Siltanen will serve as the custos.
The dissertation is available in electronic form in the University of Helsinki’s open repository Helda.
Photo: University of Helsinki / Ida Pimenoff
