About me
I am an ELLIS PhD student in Computer Science at the University of Copenhagen and a member of the FoRML group, supervised by Amartya Sanyal and Michal Valko.
My research interests lie in the theoretical foundations of modern trustworthy machine learning.
Previously, I worked on the development of a non-invasive glucose monitoring device at Liom and on improving supply chain operations at Chanel using machine learning.
Prior to that, I obtained an MSc in Statistics from ETH Zurich with a thesis on Meta Feature Learning under the supervision of Andreas Krause, and a BSc in Applied Mathematics from the Polytechnic University of Turin, where I was supported by the Young Talents scholarship.
Publications
- G. Racca, M. Valko, and A. Sanyal. “Language Generation with Replay: A Learning-Theoretic View of Model Collapse.” International Conference on Machine Learning (ICML), 2026. [arXiv]
Teaching and Supervision
I have served as a teaching assistant for the following courses:- Machine Learning B, University of Copenhagen, 2026
- Introduction to Machine Learning, ETH Zurich, 2023
- Analysis of Variance and Experimental Design, ETH Zurich, 2022
- Lukas Abild: BSc thesis on language generation in the limit, 2026
- Lukas Abild: semester project on language generation in the limit, 2025
News
- Jul. 2026: Attended ICML 2026 in Seoul and presented “Language Generation with Replay”.
- Jul. 2026: Oral presentation on “Language Generation with Replay” at LTSS Workshop in Copenhagen.
- May 2026: “Language Generation with Replay” accepted to ICML 2026 in Seoul.
- Mar. 2026: New preprint available on arXiv titled “Language Generation with Replay: A Learning-Theoretic View of Model Collapse”.
- Sep. 2025: Moved to Copenhagen to start my PhD.