Host Institution: U-GRAZ
Duration: 36 Months
Contact person: Kristian Bredies
Objectives: Development of algebraic-reconstruction-technique-type (ART) methods for the solution of inverse problems and training of neural networks with regularization terms. Establishment of convergence, and application to inverse problems and machine learning.
I got my MSc in Mathematics at the University of Pavia in September 2020. In my master thesis, in collaboration with the MOKAPLAN (INRIA) team in Paris, I discussed theoretical and computational aspects of Wasserstein medians, which represent a novel and generalized notion of median for complex data with a high potential for applications, especially when employed for statistical purposes. This is the result of a highly stimulating and fruitful journey into the field of convex optimization that started with my Bachelor thesis and developed thanks to many inspiring lectures mainly on Calculus of Variations, Functional Analysis, Optimization and Data Science.
Motivation: During my studies, I learned the importance of building a bridge between applied and abstract mathematics: two rich, strongly interconnected, but still not too communicating, worlds. This PhD program enables me to tackle modern real-world problems with multidisciplinary and innovative techniques under the guidance of motivating and qualified mentors.
Hobbies: I love to stay up to date on political and cultural events, reading and playing basketball.